3.1.7 The AMS and SMS algorithm will be updated when new or better data is obtained and whilst this procedure will occur on a regular basis the Committee reserves the right to vary measurement systems if the need arises.
HHB The largest headboard top width for the largest headsail. Measured fore and aft at right angles from the luff extension to the aft leech extension of the sail at the widest point of the headboard or head of the sail. Refer to rule 3.5.3. and Fig 6. Also refer to HWH and HWM below.
MHB The mainsail headboard measurement is the measurement fore and aft at right angles from the luff of the mainsail to the leech of the sail or its extension at the widest point of the headboard. MHB may not exceed 0.75m. See Figure 6 and refer to HWM above.
MHW Mainsail mid width measurement (measurement point is determined by folding the head of the sail to the clew and marking the leech at the fold point). Measure the girth from the leech fold point to the nearest point of the luff including any bolt rope.
MTW The mainsail three quarter (Â) width measurement point is obtained by folding the head of the sail to the mid leech point and marking the leech at the fold point. Measure the girth from the leech fold point to the nearest point of the luff including any bolt rope.
For square top mainsails the MUW measurement fold point may be located above the leech extension point used to measure HWM. In this case the MUW must be included as 0 (Zero) on the input form. Refer to Figures 7a and 7b. In this case MUW must be included as 0 (Zero) on the input form. Refer to Figures 7a and 7b.
Mercury can be taken up into fish from food via the alimentary tract; the otherroutes are through the gills and skin. Absorption from the alimentary tract has proved tobe of the greatest importance in methyl mercury accumulation; evidence for this has beenprovided by the results of investigation at sites in the drainage area of the Berounka Riverin Central Bohemia. The total mercury content in the flesh of fish from these localities isabout 10 times that recorded in their food. This coefficient of bioaccumulation can becompared with the food efficiency coeficient of fish living in open waters and feeding onthe aquatic invertebrates. Of the other aquatic organisms in the drainage area of theBerounka River, the greatest mercury levels were recorded in leeches and this can beascribed to their exclusively predatory mode of feeding. With their wide distribution indifferent types of waters, leeches (e.g. Helobdella stagnalis) may be considered as goodindicators of mercury contamination of the aquatic medium.
Download Leech 3.1.7 for Mac from our software library for free. The most popular versions among the program users are 2.2, 2.1 and 2.0. Leech can be installed on Mac OS X 10.7 or later. Our antivirus analysis shows that this Mac download is safe.
This Mac application is a product of Many Tricks. Commonly, this application's installer has the following filenames: leech220.dmg, leech221.dmg and setup.bz2 etc. The bundle id for Leech for Mac is com.manytricks.Leech. The size of the latest setup package available for download is 4.4 MB. The program lies within Internet & Network Tools, more precisely Download Managers.
Using leech mechanics for life and ES, and increasing total ES leech recovery per second where we can sustain the ES in place - meaning our EHP is actually there for us (and the ES isn't being stripped away immediately.)
When I say single target, I mean peak boss DPS with leech, exposure, curses, and convergence. Mapping DPS (mobs) is lower but they have a lower Life and Ailment threshold, so this DPS figure is what you expect to get against uniques/bosses. You can check what your mapping DPS is against mobs by unticking Convergence and disabling Conductivity.
ALSO NOTE: At 81% lightning res, 7.2 hits per second (bad rolled rings), and Glorious Vanity, ES will need 342 leech per second to sustain and not shred apart. Less leech is required with Immortal Call active.
The temperature of the blackbody is not necessarily the same forenstatite and forsterite. In determining the best fit, we variedthe temperature in steps of 5 K. The resulting spectra wereseparately scaled to fit the spectrum. This scaling factor isrelated to the mass of the dust species. The absolute massesrequires knowledge of the distances to the stars but, for each source the masses of the different dust components can be directly compared.The mineralmass ratios determined in this paper assume that they have thesame grain size and shape distribution (both around stars and inthe laboratory samples). The best fits were determined by eye andno method has been applied. This method is of sufficientaccuracy given the current quality of the lab data and given thefact that several prominent dust features still lackidentification, thus strongly affecting any method. We foundthat the temperature and mass for forsterite couldbe determined using the 23 and 33 micron complexes, whilethe enstatite values are mainly based on the 28and 40 micron features.The results of this simple fitting procedure are shown inFigs. 5 to 16 and thederived temperatures and mass ratios are given inTable 1. We also derived an estimate for thetypical temperature of the underlying continuum. For this we assumed that the continuum is caused bysmall grains with optical constants based on theamorphous silicate set 1 of Ossenkopf et al. (1992) and a continuous distribution of ellipsoids as shape distribution. We fittedthe continuum to the original, not the continuum subtracted,spectra. An independent fit based on a emissivity law gave similar temperatures. This gaveus confidence that the continuum temperature is reasonably well determined in this way. It should be noted that other shape distributions(e.g. spheres) and other sets of optical constants of amorphous olivines caneasily change the derived temperature by 20 K, more often to higher than to lower temperatures. From these fits we could in principle derive a relative mass, like in the case of enstatite and forsterite. Although the uncertainties in the (mass) absorption coefficients (due to shape, size and compositional differences) are systematic, the spread in values makes itvery difficult to interpret them and to compare them with other observations. Therefore, we have not given an amorphous over crystalline silicate mass ratio.However, since the differences between the different datasets are systematic, trends can still be derived from these numbers.For the remainder of this paper we will take thetemperature derived by the fit with the Ossenkopf data set as thecontinuum temperature (Table 1), because these fits tend to produce the best fits. We compared the temperaturesfound in this study with those found by Molster et al. (1999b, 2001a), and we found a reasonable agreement. Difference in thetemperatures found could often be described to the use ofdifferent laboratory data sets.Our simple model, consisting of only two crystalline dustcomponents and a single temperature for each dust component, fits most stars very well, see e.g. MWC922(Fig. 8). Still, it is clear that this simplemodel is not sufficient to explain all the features. The maindiscrepancies between our model fits and the ISO data lie at thewavelengths below 20 m. We note that the three stars with acontinuum temperature above 200 K all show crystalline silicatesin emission in the 10 micron region. The temperature of thecrystalline silicates has been determined based on bands atwavelengths longwards of 20 micron. These bands are dominated by cooldust, and the derived low temperatures (Table 1)are too low to explain the strength of the crystalline silicatebands in the 10 micron complex. A second, much warmer, componentmust be introduced to explain these 10 micron bands. Likely atemperature gradient is present in these sources. The discrepanciesshortwards of m do not solely reflect the presenceof a temperature gradient in these sources, but indicatethat still other dust components must be present. The 18 micron complex is badly fitted. The modelled 19.5 micron feature(originating from both forsterite and enstatite) is often muchtoo strong and the modelled 18.0 and 18.9 micron features areoften too weak when compared to the ISO spectra. The too strong19.5 micron feature might be a radiative transfer effect, sincethis feature is less of a problem in the full radiative transfermodelling (see e.g. Molster et al. 1999b, 2001a). This might indicatethat our assumption of is not correct at wavelengthsaround 19.5 m. The poor fit of the 18.0 and 18.9 micronfeatures suggests the presence of another dust component.There is more evidence for the presence of an extra dustcomponent. The 29.6 and 30.6 micron features also need extraemissivity, as is very clear in the spectra of NGC6537(Fig. 6) and of NGC6302(Fig. 7). In these two sources the 40.5 micron feature is not well fitted, suggesting that the same dustcomponent which is responsible for the 29.6 and 30.6 micronfeatures also has a peak around 40.5 m.A possible candidate for this extra dust component is diopside(MgCaSi2O6), which peaks at the required wavelengths.However this material also produces strong peaks at other wavelengths,e.g. at 20.6, 25.1 and 32.1 m, which are observed in the ISO data, but often not as strong as expected.Therefore, the identification of the carrier of the 29.6 and 30.6 micronfeatures remains open. It should be noted, that the temperature and relative mass of enstatite are estimated from the 28 and 40 micron complexes. Therefore a significant contribution of an unknown dust component to one (or both) of these 2 complexes can change the estimated temperature and abundance of enstatite.The 33.0 micron feature is not well fitted, but this feature islikely to be influenced by instrumental behaviour (see Paper II).In the 35 micron plateau we clearly miss intensity around 34.8 m in all sources. The predicted 69.0 micron feature is oftentoo weak with respect to the ISO spectra (see e.g.Fig. 15). This may be an indication for thepresence of colder dust, and thus for a temperature gradient, wewill come back to this later.Apart from all the features that are missing, we also have a problem withtoo much intensity predicted by our modelling around 27 m.This excess is mainly due to enstatite, but also forsterite contributesslightly. We are still looking for an explanation of this phenomenon.Finally, we did not attempt to fit the absorption profiles. As statedabove we assumed the dust was optically thin.Also, no attempt was done to fitthe carbon dust features, which are present in some sources.3.1 The sample starsHere we describe the model fits to the spectra ofthe individual stars, which where analyzed in Paper I.For a description of theindividual stars in this sample we refer to Paper I. From thissample we rejected Roberts 22 and VY 2-2, because the ISOsatellite was unfortunately offset when observing these twoobjects resulting in large flux jumps in the overall spectrum.This made it impossible to derive temperature estimates of thedust around these two stars. OH26.5+0.6 has also not been fitted,because below 30 m it has an absorption spectrum (Sylvesteret al. 1999), which could not be described with our simple model.The main uncertainties in the model fits are due to uncertainties inthe continuum subtraction. This leads to errors in the temperature of theorder of 10 K and mass uncertainties of the order of a factor 2.We note that for our modelling we completely rely on the laboratory datainput. This may result in systematic effects on our derived temperatures andmasses. Figure 5:A fit (dotted line) to the continuum subtracted spectrum (solid line) of IRAS09425-6040. K and K.Open with DEXTER 3.1.1 IRAS09425-6050The fit to the spectrum of IRAS09425-6040 is shown in Fig. 5.The model fit also produces a somewhat too strong 19.5 micron feature.It should be noted that the full radiative transfer calculations ofMolster et al. (2001a) produces excellent fits to the 19.5 micron feature.The broad feature at 11 m is due to SiC. This very simple modelpredicts no significant flux in the 10 micron complex due tocrystalline silicates, which is consistent with its absence in the ISOspectrum.The forsterite grains have a temperature of 85 K.This temperature agrees with the temperature range presented in the detailedradiative transfer model (Molster et al. 2001a). However, in contrast to the results presented here these detailed calculations predict that enstatite is much cooler than forsterite. As a result those models could not reproduce the relative strengthof the observed 28 and 40 micron complexes.It also resulted in an unrealistically high mass for the enstatite.Molster et al. (2001a) argue that this might have to do with the not well knownabsorptivity of crystalline enstatite. Figure 6:A fit (dotted line) to the continuum subtracted spectrum (solid line) of NGC6537. K and K.Open with DEXTER3.1.2 NGC6537The results for NGC6537 are shown in Fig. 6.The temperatures found for the forsterite (75 K)and enstatite (65 K) in NGC6537 are among the lowest found in our sample.Note that if an extra dust componentsignificantly contributes to the 40 micron complex, the temperatureof enstatite will be higher (and its mass lower)than what has been determined here.The spectral energy distribution of the complete spectrum is too broad to be fitted by a single temperature dust component. Figure 7:A fit (dotted line) to the continuum subtracted spectrum (solid line) of NGC6302. K and K.Open with DEXTER3.1.3 NGC6302The continuum subtracted spectrum of NGC6302 and its good fit areshown in Fig. 7.Molster et al. (2001b) used the same method as used in this paper, and therefore found the same temperatures. As for NGC6537, it was not possible to fit the spectral energy distribution with a single temperature dust component. Molster et al. (2001b) attribute the broad energy distribution tothe presence of a population of large grains, which mainly contribute to the long wavelength side. The presence of this population of large grainsis indicated by the gentle slope of the spectrum up to mm wavelengths (Hoare et al. 1992).The temperature found for the enstatite and forsterite, respectively65 and 70 K, are similar to the temperature of NGC6537, which in many aspects looks very similar to NGC6302. Kemper et al. (2001) assumed two temperature regimes: a cold one from 30 to 60 K, and a warm one from 100 to 118 K. Both components containforsterite and enstatite. Our results, giving a temperature somewhere in between those two regimes, is in agreement with theirs, although the exact comparison is somewhat difficult. 3.1.4 MWC922 Figure 8:A fit (dotted line) to the continuum subtracted spectrum (solid line) of MWC922. Kand K.Open with DEXTERThe fit to the continuum subtracted spectrum of MWC922 is one ofthe best we have (see Fig. 8). Especially the 40 micron complex is very wellreproduced by our model, indicating that the 50% clino- and 50%ortho-enstatite are the right proportions for this object.At m the spectrum of MWC922is dominated by PAH-features whichwere not incorporated in the fitting procedure.3.1.5 AC Her Figure 9:A fit (dotted line) to the continuum subtracted spectrum (solid line) of AC Her. K and K (dottedline). The dashed line is a 700 K (for both forsterite and enstatite)model fit.Open with DEXTERA model with cool dust fits the long wavelength part (>m)of the AC Her spectrum(dotted line in Fig. 9).However, the short wavelength features indicate the presence ofa dust component with a much higher temperature.The temperature of this material is not well constrained.In Fig. 9we show a fit of 700 K (dashed line in Fig. 9),but a similar fit could be derived with a temperature several hundreds degrees Kelvin higher or lower. Therefore it is impossible to give a reliable mass estimate for this hot component.In our modelling we only assumed a single temperature. Based on thenecessity of (at least) two different temperatures, the existence of atemperature gradient seems more likely.It is interesting to note that the overallspectrum of AC Her is very similar to that of comet Hale Bopp (Molster et al. 1999a)where we know that the dust is located in one place.Temperature differences found in the grains around this comet must thereforeoriginate from the grain size differences. Small grains canaccount for the high temperature dust emission, while bigger grains areresponsible for the low temperature dust emission.Such a scenario might also be possible for AC Her, which would imply that thedust might not have to be so close to the star as previously thought(e.g. Alcolea & Bujarrabal 1991).Jura et al. (2000) found a disk like structure for this object, whichsupports the above mentioned scenario.A full radiative transfer model fit would be necessary to completely understandthe dust distribution around AC Her, but that is beyond the scope of this paper.3.1.6 HD45677 Figure 10:A fit (dotted line) to the continuum and amorphoussilicate subtracted spectrum (solid line) of HD45677. K and K.Open with DEXTERFrom the continuum subtracted spectrum of HD45677 we first removed thebroad amorphous silicate features (Fig. 10). We cannot exclude that we also removedpart of the crystalline silicate features in the 18 micron complexin this way. This does not influence our results since these are mainly based on the 23, 28, 33 and 40 micron complexes.To fit the spectrum of HD45677 we ignored the strengthof the 19.5 micron feature, which is severely overestimated in our resulting fit. If we would have fitted the 19.5 and 40 micron features simultaneously,the 28 micron complex would have been severely underestimated. Likewise, attempts to fit the 18 and 28 micron complex together will result in a severely overestimated 40 micron complex, and also the fitsto the 23 and 33 micron complexes will become worse.It is unlikely that this discrepancy can be fully explainedby the subtraction of the amorphous silicates. Because this is not the only source with this problem, we leave this forfuture research.Malfait (1999) also studied this star. He modelled this object witha radiative transfer code. HD45677 could only be modelledwith a 2 component dust shell, consisting of a hot shell, responsiblefor the main part of the flux up to 20 m and a cool componentwhich is the main contributor to the crystalline silicates features.Due to the method we use here, our temperature estimateis based on this cool component. Malfait finds a temperature between 250and 50 K for this cool component. Unfortunately this is notspecified for the different components separately, so wecan only say that our temperatureestimates do agree with this temperature range.The predicted strength of the crystalline silicate features in the 10 microncomplex is underestimated.Since the strength of the amorphous silicate band at 10 m isuncertain, errors in the estimate of its contribution affect thestrength of the crystalline silicate bands at these wavelengths andwe did not attempt to fit the hot crystalline silicate compounds.3.1.7 89 Her Figure 11:A fit (dotted line) to the continuum subtracted spectrum (solid line) of 89 Her. K and K.Open with DEXTERBefore we fitted the continuum subtracted spectrum of 89 Her, we firstsubtracted a broad feature below the 26 to 45 m region (Fig. 11). Thisfeature is also seen in HD44179 and probably AFGL 4106and discussed in Paper II.The continuum subtracted spectrum of 89 Her is quite noisy at the longerwavelengths, which makes the fits not as well constrained as in other stars.Also in this star warmer grains are necessary to explain thecrystalline silicate structure found on top of the amorphous silicate featurein the 10 micron complex.Again, problems in the separation of the crystalline and amorphoussilicates kept us from fitting this feature.Based on the CO observations and the near-IR excess, it was arguedin Paper I, that there must be dust with different temperatures,likely a temperature gradient, around 89 Her. 3.1.8 MWC300 Figure 12:A fit (dotted line) to the continuum subtracted spectrum (solid line) of MWC300. K and K.Open with DEXTERAlthough we have argued in this paper that the strengthof the 19.5 micron feature is difficult to model correctly,we decided, because of the problems in the 28 micron complexto constrain the enstatite by the 19.5 micron feature in MWC300(see Fig. 12). If we would have fitted thestrength of the 28 micron complex, and ignored the 19.5 micron feature, wewould have derived a temperature of roughly 150 K, which is muchlarger than the forsterite temperature. It would also predictprominent features at the shorter wavelengths, which were not seen.The use of the 19.5 micron feature to constrain enstatiteresulted in a similar temperature for the forsteriteand enstatite dust species. This result, together with the reasonable fit at the 40 micron complex makes us confident in ourapproach for this star. We note however, that because of the problemswith the 28 micron feature the values for enstatite are poorly constrained.The source of the extra flux in the 28 micron feature is unknown.3.1.9 HD44179 Figure 13:A fit (dotted line) to the continuum subtracted spectrum (solid line) of HD44179. K and K.Below 15 m the spectrum is dominated by PAH features.Open with DEXTERAs for 89 Her, we removed the very broad feature in the26 to 45 m range in the continuum subtracted spectrum of HD44179.The 18 micron complex seems to containa contribution from amorphous silicates which was also removed.The result can be found in Fig. 13.It should be noted that a change in the subtraction of the broad 18 micronamorphous silicate feature, whose properties are not well determined,may change the derived temperature by more than thetypical fitting error of 10 K.The derived continuum temperature of 120 K is somewhat uncertain due tothe complex nature of the source (Waters et al. 1998).Since the short wavelength part of the continuum is formed by C-rich grains,we based our fits on the long wavelength range. This will likely underestimatethe temperature of the continuum. Taken all these uncertainties intoaccount, the crystalline silicates may have an equalor even lower temperature than the amorphous silicates, as is foundin the other stars.3.1.10 HD161796 Figure 14:A fit (dotted line) to the continuum subtracted spectrum (solid line) of HD161796. K and K.We also included crystalline water ice with a temperature of 70 K.Open with DEXTERThe 40 micron complex in HD161796 is dominated by crystalline water ice.In order to fit this spectrum (Fig. 14) we therefore added crystalline water ice to the spectrum (Smith et al. 1994). In general most features are reasonably well reproduced. The poor fit around 40 m suggests that an underlying weak broadcomponent contributes. Possibly, this is amorphous water ice.Within the errors, the crystalline and amorphous temperatures are the same.The peak wavelength of the crystalline water ice feature in our model is slightly offset from our ISO spectrum. This may be a temperature effect (Smith et al. 1994), reflecting the sensitivity of the emissivities to temperature.3.1.11 HD179821 Figure 15:A fit (dotted line) to the continuum subtracted spectrum (solidline) of HD179821. K and K.To fit the 40 and 60 micron complex, we included crystallinewater ice with a temperature of 45 K.Note, that we made no attempt to fit the amorphous silicate bands in the continuum subtracted spectrum.Open with DEXTERHD179821 shows prominent crystalline water ice features in the 40 and 60micron complex. Therefore, we also added crystalline water ice in this model fit (Fig. 15). The crystalline silicate features are rather cold, 65 K for the enstatite and75 K for the forsterite, therefore no detectable featuresare expected in the 10 micron region.3.1.12 AFGL4106 Figure 16:A fit (dotted line) to the continuum subtracted spectrum (solid line) of AFGL4106. K and K.Note, that we made no attempt to fit the amorphous silicate bands in the continuum subtracted spectrum.Open with DEXTERFor the fit to the continuum subtracted spectrum of AFGL4106(Fig. 16) we changed the ratio of clino versus orthoenstatite to 1:3. Otherwise the 44.7 micron feature would have appeared toostrong in the model spectrum. This feature is located in the wing of the 43.0 micron feature. The 43.0 micron feature is not well reproduced by our simple model, resulting in an offset of the 44.7 micron feature. A broad feature between 30 and 45 m seems present. It is not knownwhether this is the same feature as in 89 Her and HD44179. It seems to peak ata longer wavelength (38 m) than in the other two sources.Subtracting this feature, would increase both the temperature offorsterite and enstatite.The 40.5 m feature is very strong, even if one removes this broad feature,indicating that other dust species are present.No attempt has been made to fit the substructure in the 18 micron band in view of the large and uncertain amorphous silicate contribution.In this star nodetectable 10 micron structure is expected from the crystalline silicates,which is in agreement with the observations.A full radiative transfer calculation was made for this sourceby Molster et al. (1999b).The enstatite abundance derived by them was based on the (wrong) assumption that the 32.8 micron feature was due to ortho-enstatite,and is therefore difficult to compare. They also used another dataset for forsterite, which has a different intrinsic strength ratio of the 23.7 and 33.6 micron features. This results in a lower temperature for forsterite in their calculations.3.1.13 NML CygWe could not reliably fit enstatite to the spectrum of NML Cyg sincea part of the 28 micron complex was missing, due to problemswith the data reduction (see Paper I). We were able tofit forsterite to the continuum subtracted spectrum and found atemperature of about 150 K (not shown).3.1.14 IRC+10420Also for IRC+10420 we only fitted the forsterite component which appearedto be about 90 K (not shown).Enstatite could not be fitted due to the lack of data in the 28 micron region.We checked the 40 micron complex and, if enstatite would have been fitted,an ortho-enstatite to clino-enstatite ratio of 2:1 would probably give the best fit to the strength of the 44.7 micron feature. 4 CorrelationsFrom the modelling, we were able to derive values for the temperature and therelative contributions to the total amount of dust mass byforsterite and enstatite.Correlating these values can give interesting insight in thedust inventory as we will show in this section.Since in Papers I and II it was found that there is a significant difference betweenthe spectra of disk and of outflow sources, we will separate them.In all plots in this section the outflow sources are representedby a circle while the disk sources are represented by diamonds.4.1 Temperature dependence of laboratory spectra: Peak width and position Figure 17:The temperature of the forsterite grains versus the wavelengthposition of the 69 micron band. The circles are outflow sources andthe diamonds are disk sources.Open with DEXTERIn Sect. 2.1 we showed a relation between the peakposition and FWHM of the 69 micron forsterite band and the temperature of theforsterite grains. Since we have now determined the temperature of the forsterite grains, we checked this relation in our data (Fig. 17).There is no clear correlation between these two quantities.A possible explanation for this scatter behaviour might be the fact that thetemperature of the forsteriteis determined from bands in the 20 to 40 m range. The strength and widthof the 69 micron feature may be dominated by much cooler dust.The predicted strength from our simple model fits, which is lower thanthe strength in our ISO spectra (see e.g. HD179821), supports this statement. 4.2 Other temperature trendsIn Fig. 18 we compare the temperature of theenstatite and forsterite, derived from our simple model fits. For the disksources the enstatite and forsterite grains seem to have an equal temperature,while in the 3 outflow sources the forsterite seems warmer. Figure 18:The temperature of the forsterite grains versus the temperature ofthe enstatite grains. The diamonds are the disk and the circles are the outflow sources. The dashed line represents equal temperatures forthe forsterite and enstatite grains. The errors are typically in the order of10 K.Open with DEXTER Figure 19:The temperature of the forsterite grains versus the temperature of the amorphous silicate grains. The dashed line representsequal temperatures forthe forsterite and amorphous silicate grains. The diamonds correspond to thedisk sources, while the circles correspond to the outflow sources.Open with DEXTERIn Fig. 19 we compare the forsterite temperature and the amorphous silicate temperature.We first note that, in general, the crystalline forsteritegrains are colder than the underlying continuum consisting ofamorphous silicates. The difference in temperature between the amorphous and crystalline silicateshas probably to do with the difference in chemical structure.The crystalline silicates are Mg-rich (see this paper and Paper II),while the amorphous silicates must contain metals to explain their highernear-infrared absorptivity.Possibly the reaction of the Mg-rich crystalline silicates with gaseous ironmay proceed in these outflows at temperatures well below theglass temperature leading simultaneously to amorphous and dirty silicates(Tielens et al. 1998).For some stars like IRC+10420, MWC922 and IRAS09425-6040 the difference intemperature excludes the possibility of one grain population,which is partially crystalline. The crystalline and amorphous silicatesmust form two separate grain populations, which are not in thermal contact.From this simple model we cannot say that they are also spatiallyseparated, but radiative transfer modelling (Molster et al. 1999b, 2001a) indicates that this is not necessary.Finally, we note that there is no obvious separation between the disk and theoutflow sources. Because of the analogy between the different sources, weexpect that this grain segregation is valid for all sources in our sample. 4.3 Abundance trends Figure 20:The mass of the forsterite grains versus the mass of the enstatitegrains. We have assumed that they have the same size and shape distribution.The dashed line represents equal masses for the forsterite and enstatitegrains. The outflow sources are marked as a circle while the disk sources aremarked with a diamond.Open with DEXTERFrom our modelling we were also able to derivean abundance ratio for enstatite and forsterite (Fig. 20).In all sources, except NGC6302, theenstatite is more abundant than the forsterite, on average a factor 3-4.Again there seems to be a difference between the disk and outflow sources. The outflow sources have, on average, a higher enstatite over forsterite ratio than the disk sources. We note however, that the number of outflow sources is low and moredata is required to confirm this trend.If crystallization of the amorphous grains is a thermally drivenprocess, it is expected that warmer dust shells would show a highercrystalline silicate fraction. This effect would be most clear in the disks sources, wherethe dust stays at roughly the same place. In the outflow sources,the temperature of the optically thin dust shell is more an indication ofthe age of the dust. Cooler dust would be further out and therefore older,in which case an increase of the crystallinity with a lowering ofthe temperature may even be possible.We investigated the crystalline over amorphous silicatesabundance ratio with the temperature of the amorphous silicates.We excluded AC Her, 89 Her and HD45677 from this analysis because thecontinuum and feature temperatures derived likely do not reflectparticles with the same size distribution and/or location aroundthese objects, resulting in unrealistic mass ratios.No clear relation could be found between the temperature of the amorphoussilicates versus the crystallinity, neither for the disk nor for the outflowsources. The crystallization process is either a relic of the past or it isnot thermally driven.This is not too surprising. Thermally driven crystallization requirestemperatures of 1000 K (Hallenbeck & Nuth 1998), which is well above the temperature of either the crystalline or amorphous silicatesin our sources. 4.4 Crystallinity versus stellar flux ratioIt has been suggested by Waters et al. (1999) that the enstatiteover forsterite ratio increases with luminosity of the star.They based their conclusions on the ratio of the32.97 and 33.6 micron bands. They attributed the former band to enstatite,and the latter one to forsterite. Unfortunately, the 32.97 micron bandis strongly affected by instrumental effects(see Paper II), which are most prominent in thebrightest sources. These are usually also the intrinsic mostluminous sources, such as IRC+10420 and NML Cyg.Since it is likely that in high flux sources grain formationwill result in different ratios of the dust species we still decidedto investigated this scenario. We could find no trend for the stars in our sample, which had a reliable distance (and thus flux) estimate.5 DiscussionMost features can be explained with forsterite and enstatite. Howeverlaboratory spectra of these species often produce too broad bands. Mostlaboratory spectra were measured at room temperature, while the dust speciesaround our stars are in the order of 80 K. Lowering the temperature can narrowthe dust features, and this might be thenatural solution for the width problem. It can also solve another problem.The 69.0 micron feature has been attributed to forsterite. It is very sensitiveto the Fe/(Mg+Fe) ratio in olivine, and one ofthe key features for the determination of the Fe content in crystallinesilicates. However in most lab spectra it peaks at 69.7 m. Recent laboratory measurements show that this feature shifts to shorter wavelengths when the sample is cooled and will be at 69.1 m at 77 K and even at 68.8 m when measured at 4 K (Bowey et al. 2000). At the same time, the feature narrows (cf. Fig 2). So it seemsthat the temperature dependence is able to solve also this problem.From the correlation study in this chapter a clear differencebetween the disk and the outflow sources emerges. The differencesfound in Papers I and II were based on the shape and strength of thedifferent solid state features in the spectrum. These differencescan now be traced back to differences in temperature and chemicalcomposition of the circumstellar dust. The outflow sources seemmore abundant in cold enstatite grains than the disk sources,although more observations would help to quantify this better.There are also indications that in the high luminosity outflowsources the ortho- over clino-enstatite abundance ratio is largerthan unity (see AFGL4106 and IRC+10420), while in the other(disk) sources nice fits were obtained with a ratio of unity.Because of the large crystalline water-ice component in the 40 micron complex of HD161796, it is difficult to quantify itsortho- over clino-enstatite ratio. Therefore we cannot excludethat this difference we see is related to the more massive natureof the central stars of AFGL4106 and IRC+10420 rather than thedisk-outflow character.The difference between the disk and the outflow sources might be relatedto differences in conditions during the condensation of the grains outof the hotgas, and/or to differences in the conditions since the formation ofthe dust particles. Smyth (1974) found that clino-enstatiteslowely inverts to ortho-enstatite between 920 and 1220 K. Dust particleswill stay longer within this temperature range in the case of massive starsthan for low mass stars. This might explain the overabundance ofortho-enstatite in IRC+10420 and AFGL4106 with respect to the other stars.Another possible difference during the formationof these grains could be the amount of radiation pressure exerted on theparticles. It is expected to be lowerin the disk sources, otherwise the disk is likely to beblown away. Also the presence of a companion - most if not all disk sourcesare a binary system - might influence the dust forming process.If, on the other hand, we assume that the initial (forming) conditionsare similar for the disk and outflow sources, than the differences in theconditions after formation must be dominant. Time is an obviousdifferences. Also a relation with grain coagulation - the disks containlarge grains - cannot completely be ruled out.We note that clino-enstatite is often foundin meteorites on earth. The process responsible for the overabundanceof clino-enstatite in meteorites in the disks of young stars might be the sameas the one in the disks around evolved stars.We also want to note that in chondrites or primitive meteorites, i.ethe less processed ones, clino-pyroxenes are less abundant than the ortho-pyroxenes.This similarity between the disks around young and around evolved stars, isnew evidence to the hypothesis that thecircumstances and processes in disks around young stars are very similarto these in disks around evolved stars, althoughtheir origin is quite different.The equal temperature of enstatite and forsterite in the disk sources, is in principle compatible with the assumption that enstatite and forsterite are present as a composite grain in the disk sources. However, it is equally well possible that their individual equilibrium temperatures are similar, in which case they can still be individual grains. New uninterupted laboratory measurements of the optical properties of both materials, especially in the wavelength range where they absorb the stellar light and emit the thermal radiation, would allow us to make these kind of calculations and settle this point.The different temperatures found for forsterite and enstatite in the outflowsources suggests two separate grain populations in these environments.6 ConclusionsWe can summarize the main results of this study as follows:1The ISO spectra could be reasonably well fitted with laboratoryspectra of forsterite and enstatite.2The models underestimate the flux at 18, 29.6, 30.6, 48 mand sometimes at 40.5 m,which is an indication for the presence of (an)other dust component(s).No convincing identification could be made yet. Diopside does have features atmost of these wavelengths, but also strong features at others which areweak or absent in the ISO-spectra.3The 19.5 micron feature is often overestimated by our model spectra.No explanation is yet known for this phenomenon, but it should be notedthat in the full radiative transfer modelling it appeared to be much lessof a problem. This might indicate that optical depth effects play a role.Also the calculation of the absorption coefficients from theconstants instead of the absorption coefficients from laboratory particlesmight lead to differences.4The band width of the laboratory data is larger than in our ISO spectra. This difference is likely a temperature effect, andmight be used as an independent temperature indicator. Especially the69.0 micron band is very suitable for this analysis.5The temperature of the forsterite and enstatite grainsare (almost) similar for the disk sources, while the forsteriteis slightly warmer in the outflow sources. This would imply thatthe forsterite and/or enstatite grains differ slightly inthe disk and outflow sources. It is not clear whether thisdifference is due to a different formation process, or due to a differentdust process history after the grain formation.Since this trend isonly based on 3 sources more data is required to confirm the difference between the dust and outflow sources.6The crystalline silicates are colder than the amorphous silicates.This is probably due to the difference in chemical composition.No Fe is present in the crystalline silicates, while in the amorphoussilicates it is expected to explain the higher absorptivity. This differencein temperature also implies that the crystalline and amorphous grains are twodistinct grain populations.7Enstatite is on average a factor 3-4 more abundant than forsteritein our sources. There areindications that the enstatite over forsterite ratio in the outflow sourcesis higher than in the disk sources.8In the low luminosity sources the spectra were well fitted withan equal amount of ortho- and clino-enstatite, while in the two highluminosity sources more ortho-enstatite seems to be present. 9No correlation could be found between the crystallinity and thetemperature of the dust. Also the luminosity of the stars does not seem to becorrelated with the enstatite over forsterite ratio.10These simple model fits already give a good insight in the dustaround our stars. In Paper I the shape of the features naturally separated thedisk and outflow sources. In this study the differences between these twocategories became again evident. 59ce067264