Magnetic and spin evolution of isolated neutron stars with the crustal magnetic field V. Urpin1,2) and D. Konenkov2) 1) Department of Mathematics, University of Newcastle, 8 9 Newcastle upon Tyne NE1 7RU, UK 9 e-mail: [email protected] 1 2) A.F.Ioffe Institute of Physics and Technology, 194021 St.Petersburg, Russia n a e-mail: [email protected]ffe.rssi.ru J 2 2 July 1997 1 1 v Abstract 7 7 We consider the magnetic andspin evolutionof isolatedneutron starsassuming that the 0 magneticfieldisinitiallyconfinedtothecrust. Theevolutionofthecrustalfieldisdetermined 1 by the conductive properties of the crust which, in its turn, depend on the thermal history 0 8 of the neutron star. Due to this fact, a study of the magnetic field decay may be a powerful 9 diagnostic of the properties of matter in the core where the density is above the nuclear / density. We treat the evolution of neutron stars for different possible equations of state and h p cooling scenarios (standard and, so called, accelerated cooling). - Thespinevolutionisstronglyinfluencedbythebehaviourofthemagneticfield. Assuming o that the spin-down rate of the neutron star is determined by the magnetodipole radiation, r t we calculated the evolutionary tracks of isolated pulsars in the B −τ and B −P planes, s a where B and P are the magnetic field and period, respectively, and τ is the spin-down age. : Thecalculatedtracksarecomparedwithobservationaldataonthemagneticfieldandperiod v i of pulsars. This comparison allows to infer the most suitable equations of state of nuclear X matterandcoolingmodelandtodeterminetherangeofparametersoftheoriginalmagnetic r configurations of pulsars. a 1 1 Introduction Ourknowledgeofthepropertiesofmatterinneutronstarcoresisasubjectofmanyuncertainties. Thecoremayconsistofthenormalnuclearmatterwith”standard”properties,somecomponents of this matter may be in superfluid or superconductive state, and the presence of exotic phases (free quarks, pion condensate etc.) cannot be excluded a priori as well. The most direct way to study the nature of a superdense matter is a comparison of surface thermal radiation measurements with predictions of neutron star cooling models (see, for example, Van Riper 1991, Nomoto & Tsuruta 1987). Unfortunately, observations of thermal emission from neutron stars are few, and available observational data don’t allow to infer the properties of the neutron star interiors (see, e.g., O¨gelman 1994). Another way to study neutron star cores is associated with the evolution of the magnetic field. A surface strength of the magnetic field is estimated for the majority of 700 known pulsars of different ages. The evolution of the field is mainly determined by the conductive properties of plasma which, in its turn, depend on the temperature. The thermal evolution is strongly influenced by the state of matter in the central region of the star. The link between the magnetic evolution and properties of the core matter provides generally one more window into the neutron star interior. The evolution of the magnetic field, however, depends on its configuration which is unknown since the origin of the field is not clear until now. The field could have been amplified from the weak field of the progenitor star due to the magnetic flux conservation. In this case, the magneticfieldlinesoccupyprobablyasubstantialfraction oftheneutronstarvolumeandpasses through the core. It is also possible, however, that the magnetic field is confined to relatively not very deep layers. Such a crustal magnetic configuration can be generated, for example, by the thermomagnetic instability during the first years of the neutron star life (Urpin, Levshakov & Yakovlev 1986) or by convection which arises in first 10-20 seconds after the neutron star is born (Thompson & Duncan 1993). The electric currents maintaining crustal magnetic fields are anchored in the crust but, due to a dependence of the crustal conductivity on the temperature, the evolution of such configurations is also influenced by the properties of matter in the core. This effect has been first argued by Urpin & Van Riper (1993). Thepresentpaperconsidersinmoredetailtheinfluenceoftheinternalstructureofaneutron star on the decay of crustal magnetic fields. The crustal field decay has been a subject of study for a number of papers (see, e.g., Sang & Chanmugam 1987, 1990, Urpin & Muslimov 1992). These papers have been addressed, however, to the magnetic evolution of the particular model of a neutron star and did not examine the effect of equation of state and cooling scenario on the decay. In the present study, we treat the behaviour of the magnetic field for the neutron star models with different equations of state (BPS, FP, PS) and different thermal histories. The representatives of cooling scenarios are the so called ”standard” cooling (associated with the standard neutrino emissivities; see, e.g., Van Riper 1991) and ”accelerated” cooling which can be caused by the enhanced neutrino emission due to either direct URCA processes (Lattimer et al. 1991) or the presence of a pion-condensate or free quarks in the core. It turns out, that the magnetic evolution is strongly sensitive to details of the internal structure, and it provides a hope to discriminate between theoretical models by a comparison with observational data on pulsar magnetic fields. Our knowledge of the magnetic field strength of neutron stars comes mainly from radio pulsars with measured spin-down rates. The real ages of these pulsars are usually unknown but 2 one can determine from observational data the so called ”spin-down” age, τ = P/2P˙, where P is the spin period of the neutron star. Therefore, for a comparison with observational data, it is more convenient to analyse a dependence of the magnetic field on the spin-down age but not on the real age. Assuming that the spin-down torque on a pulsar is determined by its magnetic dipole radiation, we calculate the magnetic and spin evolution of neutron stars and plot the evolutionary tracks in the B −τ and B −P diagrams. A comparison of these tracks with the observable distribution of pulsars in the B−τ and B −P planes allows to determine the most suitable parameters of the initial magnetic configurations of neutron stars. The paper is organized as follows. The main equations governing the magnetic and spin evolution of neutron stars with the crustal magnetic configuration are presented in Section 2. The numerical results are described in Section 3. In Section 4, we compare the theoretical decay curves with the observable distribution of pulsars and determine the parameters of their magnetic configurations. The summary of our results is given in Section 5. 2 Basic equations We assume that the magnetic field of the neutron star has been generated in the crust by some unspecified mechanism during or shortly after a neutron star formation. The main fraction of the crust volume quickly (< 1 yr) solidifies, thus one can neglect the effect of fluid motions. In the solid crust, the evolution of the magnetic field, B~, is governed by the induction equation ∂B~ c2 1 = − ∇× ∇×B~ , (1) ∂t 4π σ (cid:18) (cid:19) where σ is the electrical conductivity. We neglect in this equation the anisotropy of the con- ductivity that is justified if ω τ < 1 where ω is the gyrofrequency of Fermi electrons and B e B τ is their relaxation time. We consider the evolution of a dipole magnetic field alone. In this e case, B~ = ∇ × A~ and the vector potential, A~, may be written in the form A~ = (0,0,A ); ϕ A = s(r,t)sinθ/r where r, θ and ϕ are the spherical coordinates. The function s(r,t) satisfies ϕ the scalar equation 4πσ∂s ∂2s 2s = − . (2) c2 ∂t ∂r2 r2 We impose the standard boundary condition for a dipole field R∂s/∂r +s = 0 at the surface r = R. For the crustal field, s should vanish in deep layers. At the late evolutionary stage, the field can diffuse in relatively deep layers of the crust and even reach the crust-core boundary. We assume that the core matter is in the superconductive state and the magnetic field does not penetrate into the core. Note that the dipole configuration is chosen only for the sake of conveniency, andhighermultipolescanbetreatedinasameway. Theirevolution isqualitatively similar to the evolution of the dipole field. The field decay is determined by the conductive properties of the crust which, in its turn, dependonthemechanismofscatteringofelectrons. Likely, scatteringonphononsandimpurities are most important in the neutron star crust; which mechanism dominates depends on the density ρ and temperature T. The total conductivity of the crust is given by −1 1 1 σ = + . (3) σph σimp! 3 The phonon conductivity, σ , is proportional to T−1 when T is above the Debye temperature ph andtoT−2 forlower T (Yakovlev & Urpin1980). Theimpurityconductivity, σ , is practically imp independent of T and its magnitude is determined by the impurity parameter Q: 1 Q = n′(Z −Z′)2 , (4) n n′ X where n′ is the number density of an interloper specie of charge Z′; n and Z are the number density and charge of the dominant background ion specie; summation is over all species. In the presentcomputations,weusetheimprovednumericalresultsforσ obtainedbyItoh,Hayashi& ph Kohyama(1993)forthesocalledequilibriumchemicalcompositionandtheanalyticalexpression for σ derived by Yakovlev & Urpin (1980). imp The conductivity depends on the crustal temperature which is determined by the thermal evolution of the neutron star. In its turn, the thermal history is determined by the properties of matter in the core where the density is above the nuclear density. Due to this, the magnetic evolution oftheneutronstarwiththecrustalmagnetic fieldisinfluencedbythestate ofinteriors and, in principle, can provide information on the properties of superdense matter. Withtheexceptionofashortinitialphase(1−103 yrdependingonthemodel)atemperature non-uniformity is small over the main fraction of the crust volume. Since the time scale of the decay of the magnetic field is typically longer than this initial phase, we assume the crust to be isothermal with T = T where T is the temperature at the bottom of the crust. c c The calculations we present here are based on cooling scenarios considered by Van Riper (1991)fordifferentneutronstarmodels. Forourpurposes,wehaveselectedthecoolingcurvesfor models with a baryon mass of 1.4M⊙ constructed with the equations of state of Pandharipande and Smith (1975; hereafter PS), Friedman and Pandharipande(1981; hereafter FP), and Baym, PethickandSutherland(1971; hereafterBPS).TheBPSmodelisrepresentativeofsoftequations of state, which result in high central densities and small mass of the crust. The PS model was chosen as a stiff model with a low central density and a massive crust. The FP equation of state is a representative intermediate model.The stiffer the equation of state, the larger are the radius and crustal thickness for a given neutron star mass. Thus, the radii are 7.35, 10.61 and 15.98 km for the BPS, FP and PS models, respectively; the corresponding crustal thicknesses are ≈ 310, 940 and 4200 m; we assume the crust bottom to be located at the density 2×1014 g/cm3. Our choice of models is not exhaustive but it illustrates well the role of the equation of state in the magnetic evolution of neutron stars. For each equation of state, we consider two substantially different cooling scenarios which represent the slowest (”standard”) and most rapid (”accelerated”) cooling models among the results given by Van Riper (1991). The standard cooling model corresponds to a star with normal npe-matter in the core and with standard neutrino emissivities. The accelerated cooling represents themodels withenhanced neutrinoemissions inthecore either duetothe directUrca processes in the npe-matter or due to the presence of quarks or pion condensate. Note that the cooling scenario is not generally independent of the equation of state. For instance, enhanced neutrino emission may result from exotic constituents of matter but these constituents can be formed at a very high density which can be reached only for soft equations of state. Due to this, the models with a stiff equation of state and accelerated cooling or with a soft equation of state and standard cooling look questionable. However, our knowledge of the phase transition from the normal nuclear matter to the exotic one is at best uncertain therefore we present here the results for all combinations of cooling scenarios and choosen equations of state. 4 The standard models have a high surface temperature (≥ 106 K) during the first ∼ (3 − 5)×105 yr. For the accelerated cooling models, the surface temperature declines abruptly at ∼ 1−103 yr (depending on the equation of state) and after that the star cools down relatively slowly. For conveniency, we plott in Fig.1 the dependenceofthe internal temperatureon theage for differentneutronstar models. Note that cooling models given byVan Riper(1991) end when T = 3×104 K, T is the surface temperature. The results of additional cooling calculations for s s ages up to 109 yr (as well as the density profiles for different models) have been kindly provided us by Prof K.Van Riper. These calculations involve balancing the heat capacity of degenerate fermions against surface radiation and extrapolation of the smooth T /T (T ) from ≈ 5 to 1. m s m However, at these low temperatures (T < 106 K), σ is dominated by impurity scattering in the region where the currents maintaining the magnetic field are concentrated, and the field decay is thus insensitive to the details of the cooling. Due to this, also different reheating mechanisms which can make the evolution at late times non-linear cannot affect appreciably the behaviour of the magnetic field. Equation (1) with the corresponding expressions for conductivities determines the depen- dence of the neutron star magnetic field on the age. However, our knowledge of the surface field strength and its behavior with time comes mainly from radio pulsars with measured spin-down rates. For the most of these pulsars, the real age is unknown and observational data provides information only on the so called ”spin-down” age, τ = P/2P˙, where P and P˙ are the spin period and spin-down rate, respectively. Therefore, for a comparison with observational data, it is more convenient to consider the dependence of B on τ and P rather than on t. With the assumption that the spin-down torque on the pulsar is determined by its magnetodipole radia- tion (Ostriker & Gunn 1969), the spin period and spin-down rate are related to the surface field strength at the magnetic equator, B , by s PP˙ =αB2 , (5) s whereα = 9.75×10−40R6/I s,R = R/106cm,I = I/1045 g·cm2,I isthemomentofinertia. 6 45 6 45 We assume that the spin and magnetic axes are perpendicular. The decay of the magnetic field decreases the spin-down rate and, hence, changes the evolutionary tracks of pulsars both in the B - P and B - τ planes. 3 Numerical results We solve the equation (2) with the corresponding boundary and initial conditions by making use of the implicit differencing scheme. The original field is assumed to be confined to the outer layers of the crust with the density ρ ≤ ρ . The calculations have been performed for a wide 0 range of ρ , 5×1013 g/cm3 ≥ ρ ≥ 1010 g/cm3. It was argued by Urpin & Muslimov (1992) 0 0 that the decay is sensitive to the initial depth penetrated by the field and, hence, to the value of ρ . However, the evolution is much less flexible to the particular form of the original field 0 distribution within the layer ρ < ρ . The only exception is the case when a substantial fraction 0 of the initial currents is mainly concentrated at ρ ≪ ρ . In the present calculations, we choose 0 s(r,0) in the form s(r,0) = (1−r2/r2)/(1−R2/r2) at r > r , (6) 0 0 0 s(r,0) = 0 at r <r , 0 5 where r is the boundary radius of the region originally occupied by the magnetic field, ρ = 0 0 ρ(r ). 0 Unfortunately, at present, thereis noplausibleestimate of theimpurity parameter, Q, in the neutron star crust. In our calculations, the value Q is taken within the range 0.1 ≥ Q ≥ 0.001, and it is assumed to be constant throughout the crust. Tosimulate therotational evolution onerequiresthevalueofthemomentofinertia, I, which is determined by the equation of state. We use for I the simple analytic expression suggested by Ravenhall & Pethick (1994) and acceptable for a wide variety of equations of state. Fig.2showstheevolutionofthesurfacemagneticfield,normalizedtoitsinitialvalue,B . The 0 top andbottom panelsrepresentthemodels withstandardand accelerated cooling, respectively. It is seen that the decay is qualitatively different for different cooling scenarios. For the slowly cooling standard models, the higher crustal temperature leads to a lower electrical conductivity and, hence, to a more rapid decrease of the field during the neutrino cooling era. The rate of dissipation of the magnetic field is particularly high during the early evolutionary stage when the conductivity is determined by electron-phonon scattering in the region in which currents are concentrated. During this short stage (≤ 1 Myr), the surface field strength can weaken by a factor of 5-1000, depending on the original depth penetrated by the field and the equation of state. The stiffer is the equation of state, the slower is the field decay at a fixed ρ . This 0 dependence is evident because the rate of dissipation is determined by the length scale of the field which is larger for the neutron star with a stiffer equation of state. Note that the difference in the field strength may be very large after the initial stage (t ≤ 1 Myr) for the models with different equations of state even if the original field is confined to the layers with the same ρ . 0 Thus, the field confined originally to the density ρ = 1013 g/cm3 decreases by a factor of ∼ 5, 33 and 100 after 1 Myr for the PS, FP and BPS models, respectively. Between 0.1and2Myr,thedominantconductivitymechanismchangesfromelectron-phonon to electron-impurity scattering in the standard cooling model. As a result, the conductivity increases and the rate of field decay slows down. The characteristic feature of the models with standardcoolingisthepresenceofflatportionsofthedecaycurvesatt ∼ 1−300Myrdepending on the impurity parameter. The lower is the impurity content, the longer is the plateau on the corresponding decay curve. These plateaus reflect a change of the conductivity regime in which σ becomes relatively high and independentof T. The length of plateau dependson the equation of state at given Q. Thus, at Q = 0.001, the field is practically constant during ∼ 3 × 107, 2×108, 3×109 yr after the initial stage for the BPS, FP and PS models, respectively. During the impurity dominating stage, the decay may be extremely slow if the impurity content is low. Note that the decay follows approximately a power law during the late evolutionary stage when the neutron star leaves a plateau. This simple dependence can be obtained from the analytical consideration of diffusion of the magnetic field in the crust (see Urpin, Chanmugam & Sang 1994). A power law decay lasts until the magnetic field reaches the crust-core boundary. After this point, the decay becomes faster. Departures from the power law are especially pronounced fortheBPSmodelafter108 yr,sincetheradiusissmallestandthediffusiontime-scaleisshortest for this model. For other models, diffusion throughout the crust proceeds on a time scale longer than 109 yr, thus departures are negligible at t ≤109 yr. The accelerated cooling models result in substantially different field decay. Most noticeable is the slow decay - depending on the original depth and the equation of state, the field weakens by a factor 1.5-100 after 109 yr. This slow decay is due to the higher conductivity in the cooler crust. The decay curves for the accelerated cooling models do not practically exhibit the 6 plateaus as do the curves for the standard models. The only exception is the case when the original field is confined to the layers with a small depth (ρ ≤ 1011 g/cm3) but even for such 0 magnetic configurations the plateaus are much less pronounced than in the top panels. This is due to the fact that the internal temperature falls down very quickly for the neutron star with accelerated cooling. In reality, impurity scattering dominates the conductivity in the deep layers practically from the beginning and those magnetic configurations which are anchored in these layers do not experience a change of the conductivity regime. The decay of such a deeply anchored field is monotonic. However, inspite of a low temperature of the accelerated cooling models, the conductivity can be mainly determined by electron-phonon scattering in the layers with a relatively low density (ρ ≤ 1011 g/cm3) during the initial evolutionary stage. If the field is originally confined to the layers with such a low density, the decay curves can exhibit the plateaus but, of course, they lie much above the corresponding plateaus for the standard cooling models since the crustal conductivity is much higher and the decay is much slower for the rapidly cooling models. Figure 3 shows the strength of thesurfacemagnetic field at the poleversus thechatacteristic age, τ = P/2P˙. For all models, the calculations have been done for the initial polar magnetic field B = 3 × 1013 G and the initial period P = 0.01 s. Note that the behaviour of tracks o 0 in the B-τ plane are sensitive to the initial field strength since the spin evolution depends non- linearly on the magnetic field. On the contrary, the decay curves in Fig.2 do not depend on B 0 because the magnetic evolution is determined by the linear equation (2). The dependences of B on τ are qualitatively similar to the decay curves: the standard cooling models exhibit the plateaus whereas the accelerated cooling models show a much more monotonic behaviour. The tracks are weakly sensitive to the initial spin period P : the neutron star forgets about its initial 0 rotation after a relatively short time (of course, if P is not originally very large). The magnetic 0 evolution in terms of thespin-downage τ proceedsslower than inareal timebecauseonealways has τ(t) > t for a decreasing magnetic field. Thus, the tracks for the BPS model with standard cooling reach the plateaus after τ ∼ 3 − 100 Myr depending on the initial depth penetrated by the field. These spin-down ages correspond to t ∼ 1 Myr. Obviously, the difference is less pronounced for the FP and PS models which experience a slower decay. 4 Discussion A comparison of the computed tracks with the available observational data on pulsar magnetic field allows, in principle, to infer some parameters of their magnetic configurations. Note, however, that the interpretation of pulsar data is a sublect of many uncertainties, and there is no commonly accepted point of view which behaviour of the magnetic field is most suitable to account for a great variety of observations. A some evidence of a relatively slow field decay has been obtained recently by making use of the so called method of population synthesis (Bhattacharya et al. 1992, Wakatsuki et al. 1992, Hartman et al. 1996). In this method, one assumes distributions for the initial pulsar parameters, and laws governing their evolution, to simulate a population of radiopulsars. The main advantage of the population synthesis method is to model the selection effects in detail (Lorimer et al. 1993). Comparing the simulated population with the observed one, Bhattacharya et al. (1992) and Hartman et al. (1996) concluded that models in which the magneticfielddecayslittleduringtheactivelifetimeofaradiopulsargivethebestdescriptionof 7 the observations. This conclusion seems to be weakly sensitive to many assumptions concerning the initial pulsar properties and their evolution. For example, Hartman et al. (1996) used the improved information that has become available since the paper by Bhattacharya et al. (1992) was written, on the distribution of electrons, on the velocity distribution of newly born neutron stars and on the effect of gradient in the birthrate in the Galaxy. In spite of a large difference in these input parameters, the authors of the both papers concluded that the models with long decay times (≥ 30 Myr) give acceptable fits to the observational data. According to Hartman et al. (1996), good fits can be obtained if the mean magnetic field of the pulsar population ranges from2×1012 to 4×1012 G. Note thattheseconclusions hasbeenobtained for theparticular case of the exponential fielddecay which appearsto berather questionable from thetheoretical point of view. Besides, the exponential decay leads to a very fast decrease of the field for ages longer than the decay time-scale. Due to this, the exponential law can be satisfactory for making old pulsars with relatively weak magnetic fields (∼ 1011 G) in sufficient numbers only if the decay time is comparable with the true age of oldest pulsars. The situation may be quite different for other decay laws which are characterized by a slower decrease than the exponent. Note also that the recent analyses of slowly rotating pulsars (see Han 1997) indicates some problems in the evolution of these objects if the fields decay exponentially. Probably, one of disadvantages of the population synthesis method concerns very young pulsars with t ≤ 105 yr. The number of these pulsars is small and, correspondingly, they give a negligible contribution to any statistical analysis. The behaviour of newly born pulsars during first 105 yr of their life, however, may be very important for our understanding of the neutron star physics. It has been argued by Lyne (1994) that the magnetic fields of pulsars in supernova remnants are appreciably stronger than the average field of the main population. The number of these youngest known pulsars is relatively small and does not allow to make a statistically reliable conclusion at present. However, if this pointis correct then either pulsars in supernova remnants are not progenitors of the standard pulsar population or the magnetic field has to decay. Note once more an impotance of selection effects for any analysis of the pulsar population. In the case of supernova remnants, these effects favour pulsars with high magnetic fields because they are bright and can be detected agains the strong emission from the shell and plerion. Also, many pulsars in supernova remnants are discovered accidentally rather than systematically, and many could have been missed. In Fig.4, we plot the magnetic field versus the spin-down age τ for 440 pulsars taken from the paper by Taylor, Manchester & Lyne (1993). These data have been completed by the most recent data on pulsars in supernova remnants published by Frail, Goss & Whiteoak (1994). We did not plot in this figure binary pulsars, pulsars in globular clusters and those pulsars which has the magnetic field < 1010 G. Note that almost all these low-magnetized pulsars (with the exception of two) either enter binary systems or have a period shorter than 100 ms and may be related to millisecond pulsars. The magnetic and spin evolution of all excluded objects may be essentially influenced by mass transfer and may differ from that of isolated pulsars. Pulsars in supernova remnants are marked by starlets. We calculated the polar strength of the magnetic field from P and P˙ assuming the FP equation of state. Fig.4 shows a clear decrease in the surface field strength with increasing τ. However, as pointed out by Lyne, Ritchings & Smith (1975), an appreciable portion of the above trend may be caused by a much larger range in P˙ than in P of the observed pulsars, thus the B −τ distribution cannot be a reliable argument for the field decay. Nevertheless, the characteristic magnetic field of very young pulsars can be well inferred from this figure. A tendency of youngest pulsars to have the magnetic field 8 ∼ (1−2)×1013 G is clearly seen from the plot. This value is much higher than the average field required for good fits in population synthesis method. In our mind, both the above points can be easily understood in the frame of the crustal magnetic field model. The decay scenarios for neutron star models with standard cooling give a good compromise between the results of Lyne (1994), on one hand, and Bhattacharya et al. (1992) and Hartman et al. (1996), on the other hand. Actually, the decay curves for standard cooling show a relatively rapid decrease of the magnetic field at the beginning of the evolution, at t ≤ 105 yr (see Fig.2), for any equation of state. During this short initial phase, the field strength can be reduced from the value ∼ (6−30)×1012 G, typical for pulsars in supernova remnants, to the ”standard” pulsar field ∼ (1−3)×1012 G. Pulsars with such short ages will not evidently contribute to the statistical properties of the pulsar population because of a small number of such young objects. After the initial phase, at t > 105 yr, the decay of the crustal field slows down and the field can be practically unchanged during a long time. The age of the great majority of known pulsars lies within the time interval when decay curves show the plateaus, thus, our theory provides a natural scenario of a slow decay for these pulsars. When comparing the simulated population obtained in population synthesis with the observed one, these pulsars give the dominating contribution. The parameters most suitable for the initial magnetic configurations of neutron stars and their crust can be easily estimated if one adopts the above scenario. As it was mentioned, the magnetic field of young pulsars in supernova remnants is usually by a factor 3-10 higher than the average pulsar field. Since the field strength on the plateaus in Fig.2 is determined by the initial depth penetrated by the field, one can estimate ρ which corresponds to such a decrease 0 during the initial phase. This density is approximately (1.5 −2) × 1014, (5 −10) × 1013 and (1 −10) ×1012 g/cm3 for the BPS, FP and PS models, respectively. The initial field should occupy practically the whole crust in the case of the BPS model, however, the fraction of the occupied crust volume is much smaller for the FP and PS models. The corresponding depth ranges from ≈ 550 to ≈ 650 m for the FP model and from ≈ 900 to ≈ 1100 m for the PS model. The length of plateaus on the decay curves (see Fig.2) is determined by the impurity pa- rameter, Q. The population synthesis method gives acceptable fits for observations if the mag- netic field decays little on a time scale comparable with the active life time of a radio pulsar, 100 ≥ t ≥ 30 Myr (see Bhattacharya et al. 1992, Hartman et al. 1996). Practically for all con- sidered equations of state, theplateaus go until t ∼ 30−100 Myr if Q ≈ 0.01−0.03. If thedecay time acceptable for good fits will turn out to be shorter (for example, for other decay laws), then the impurity parameter may be even larger. Unfortunately, at present there is no plausible ideas regarding the impurity content in the neutron star crust, however, the value Q ∼ 0.01 seems to be consistent with the generally used estimate for this parameter, Q ∼ 0.1−0.001, and corresponds to an intermediately polluted crust. The models with accelerated cooling seem to be less attractive for an interpretation of the abovedataonpulsarmagneticfields. Thedecayduringtheinitialphaseismuchlesspronounced for these models and the initial phase is longer than in the case of standard cooling. In Fig.5 we show the evolutionary tracks in the B −P plane for different initial conditions and equations of state assuming the standard cooling scenario. We also plot in this figure the observed distributionof radiopulsars. Note that theobserved positions of radiopulsarsdepends on the equation of state since the magnetic field is calculated from the data on P and P˙ by making use of equation (5) which depends on the moment of inertia. The latter, in its turn, is determined by the equation of state. This is the reason why the measured magnetic fields are 9 different for the BPS, FP and PS models in Fig.5. We choose the parameters of tracks with the only wish to illustrate how the pulsars with different original magnetic configurations move in the B −P plane and to show that the observed variety of isolated pulsars with a wide range of measured magnetic fields and periods can be naturally formed in the frame of our model. It turns out, however, that the hypothesis of the crustal magnetic field is not satisfactory for the neutron star models based on soft equations of state. For the BPS model, it is rather difficult to account for the existance of pulsars with a high magnetic field, B ∼ (3−5)×1013 G, and with a long period, P ∼ (2 − 5) s for any reasonable choice of the initial parameters. Such pulsars cannot be formed in the course of evolution even if the original magnetic field occupies the whole crust (ρ = 2×1014 g/cm3) and is as high as 1014 G. The main reason of this is small 0 radius and thickness of the crust of soft models and, hence, a relatively fast decay of the crustal magnetic field. Due to a fast decay, the energy loss for magnetodipole radiation decreases very rapidly and the neutron star cannot slow down to long periods ∼ 2−5 s while having a high magnetic field. This problem concerns all models with the equation of state softer than that of Friedman-Pandharipande. For stiffer equations of state, the situation is more optimistic and the observational data can be well accounted for the hypothesis of the crustal origin of the magnetic field. The radius and thickness of the crust is larger for stiffer models and, therefore, the field decay is much slower. In contrast to a widely accepted opinion that any model with the crustal magnetic field should have troubles in the explanation of the evolutionary history of relatively old pulsars with strong magnetic fields, it turns out that models with stiff equations of state are quite suitable for this. Even the the existence of the oldest among highly magnetized pulsars can be easily understood in the frame of the crustal model. This can be well illustrated by the curve 3 in the PS(s) panel of Fig.2. This decay curve corresponds to the magnetic configuration which occupies initially only ∼ 1/4 of the thickness of the crust, however, the decay is already extremely slow in this case: the field is reduced by a factor ∼ 10 after 1010 yr. Therefore, there is nothing particular for the neutron star to have the magnetic field ∼ 1012 after 1010 yr of the evolution if the initial field strength was more or less standard, ∼ 1013 G. For instance, the famous pulsar in Her X-1 which has the magnetic field of the order of (1−3)×1012 G and is about 600 Myr old (Verbunt, Wijers & Burm 1990) requires even a less strong initial field, or its initial field can be confined to less dense layers. Note that all considered models have no troubles in making a sufficient number of weakly magnetized and shortly periodicisolated pulsars. Such pulsarscan beformedif either theinitial depth penetrated by the field is smaller than for the mean pulsars, or the parameter Q is a bit larger, or the initial field strength is weaker (see Fig.5). The parameters of the initial magnetic configurations required for a description of observed pulsars are more or less standard. For example, the initial strength of the magnetic field can probably lie within the range 3×1013 ≥B ≥ 1012 G for the majority of pulsars if the equation 0 of state is close to that of Pandharipande-Smith. The initial depth penetrated by the magnetic field ranges typically from 1012 to 1013 g/cm3 for the main pulsar population and only very few pulsars requires the original field anchored in bit deeper layers. This conclusion seems to be in a good agreement with the estimate of ρ obtained above. Obviously, the original field has to 0 be anchored in more deep layers for models with softer equations of state. The FP model, for instance, gives an acceptable description of the pulsar population if the field occupies originally the crustal layers with ρ ∼ 1013 − 1014 g/cm3. Correspondingly, the initial field should be 0 stronger for this model, 6 × 1013 ≥ B ≥ 3 × 1012 G. The value of the impurity parameter 0 10