A Theoretical Investigation of He i Line Profiles for the Spectroscopic Analysis of DB White Dwarfs Patrick Tremblay, Pierre Bergeron, Alain Beauchamp Abstract We present a comprehensive investigation of He i line profile calculations used in the spectroscopic analyses of DB white dwarfs. Our study includes an in-depth photometric and spectroscopic analysis of all DB white dwarfs in the Data Release 17 of the Sloan Digital Sky Survey, examining the effects of frequency sampling, Doppler broadening, line dissolution, broadening by neutral particles, and 3D hydrodynamical corrections on our results. More importantly, we compare the outcomes obtained from the semi-analytical He i Stark profiles commonly used in DB white dwarf spectroscopic analyses with our recent calculations of Stark-broadened profiles derived from computer simulations. show]patrick@astro.umontreal.ca show]bergeron@astro.umontreal.ca I Introduction There are two primary methods for measuring the stellar parameters of white dwarf stars. The first, known as the spectroscopic method ( 12 ; 25 ; 46 ; 13 ) , involves comparing observed spectra with predictions from synthetic models to determine the effective temperature ( T_{\rm eff} ) and surface gravity ( \log\textsl{{g}} ) of each object, as well as its atmospheric composition, including any trace elements. The resulting T_{\rm eff} and \log\textsl{{g}} values are then converted into stellar mass ( M ) using evolutionary sequences for white dwarfs (e.g., 26 , 7 ), which provide the necessary mass-radius relations. An alternative approach, known as the photometric method ( 11 ; 10 ; 9 ; 31 ) , relies on photometric observations (magnitudes), which are converted into averaged fluxes using appropriate photometric calibrations ( 38 ) . These fluxes are then compared with synthetic flux predictions, averaged over each bandpass. In this method, the two fitted parameters are T_{\rm eff} and the solid angle \pi(R/D)^{2} , where R is the white dwarf’s stellar radius, and D is its distance from Earth. The distance may be determined directly from trigonometric parallax measurements, or alternatively, a specific value for the white dwarf mass (or surface gravity) can be assumed. Once again, the T_{\rm eff} and R values can be converted into mass using white dwarf mass-radius relations. In certain cases, a hybrid spectroscopic and photometric approach is applied, such as for DZ and DQ stars, where the photometric method is used in conjunction with atmospheric composition constraints derived from spectroscopic fits to optical spectra ( 23 ; 24 ) . Both methods have been successfully applied for the past 30 years or so (see 8 for a review), although the number of objects available for analysis was initially quite limited. For example, 12 employed the spectroscopic method to analyze a sample of only 129 DA stars in order to determine their mass distribution. Likewise, 10 used the photometric method on a sample of 152 cool white dwarfs, utilizing optical BVRI and infrared JHK photometry as well as published trigonometric parallax measurements. At the time, their relatively small sample represented nearly all the parallax measurements available in the literature. Today, however, thanks to large spectroscopic surveys like the Sloan Digital Sky Survey (SDSS, 60 ), extensive photometric surveys like SDSS and the Panoramic Survey Telescope and Rapid Response System (Pan-STARRS, 17 ), and most importantly, the Gaia astrometric survey ( 27 ) , spectroscopic and photometric analyses of white dwarf stars have advanced significantly, allowing thousands of objects to be analyzed simultaneously ( 9 ; 41 ; 47 ; 16 ; 49 ) . These studies have, for the first time, enabled a detailed comparison of the spectroscopic and photometric methods for determining the stellar parameters of white dwarfs, allowing for assessments of their validity, accuracy, and precision. Such in-depth comparisons were carried out for DA and DB white dwarfs by 29 and 54 . Although the measurements generally agree, significant discrepancies were noted, potentially due to issues with photometric calibrations or limitations in the model atmospheres, particularly in line-broadening theory and the treatment of convective energy transport. In this paper, we focus on DB white dwarfs, whose spectra are dominated by neutral helium lines, and may also contain weak hydrogen lines (primarily \rm{H}{\alpha} ) in the case of DBA stars. 30 presented what is arguably the most extensive and comprehensive photometric and spectroscopic study of the DB white dwarf population to date, analyzing ugriz photometry and medium-resolution spectroscopy from the SDSS, combined with trigonometric parallax measurements from the Gaia mission. The authors conducted a detailed comparison of the T_{\rm eff} , \log\textsl{{g}} , and mass measurements derived from both photometric and spectroscopic techniques to evaluate the precision and accuracy of each method. The most relevant findings for our discussion are shown in their Figures 7 and 10, which display the \log\textsl{{g}} and mass distributions, respectively, as functions of T_{\rm eff} for both methods. The photometric distributions exhibit the expected mean values of \langle\log\textsl{{g}}\rangle\sim 8 and \langle M\rangle\sim 0.6 M_{\odot} across the entire temperature range, while the spectroscopic distributions show considerable deviations from these means at most temperatures. Several explanations were proposed for these discrepancies. For instance, above T_{\rm eff}\sim 27,000 K, where spectroscopic masses appear too low, residual flux calibration issues with the SDSS spectra were suggested as a possible cause. This issue had also been reported by 28 , who observed a similar phenomenon in the spectroscopic \log\textsl{{g}} distribution of DA white dwarfs from the SDSS. At lower temperatures ( 22,000~{\rm K}>T_{\rm eff}>16,000 K), the spectroscopic \log\textsl{{g}} values for DB stars appear overestimated relative to the mean. This issue could be partially mitigated by applying the 3D hydrodynamical corrections of 19 , who developed 3D model atmospheres for pure helium-atmosphere white dwarfs using the CO 5 BOLD radiation-hydrodynamics code. The corrected \log\textsl{{g}} values, shown in the bottom panel of Figure 7 in 30 , show improvement down to around 19,000 K; however, at lower temperatures, the spectroscopic \log\textsl{{g}} values appear significantly underestimated, suggesting that the 3D corrections may have been overestimated. It was noted, however, that these 3D corrections were initially available only for pure helium atmospheres, and that calculations including hydrogen could potentially resolve this issue. More recently, 20 revisited this issue by calculating a comprehensive set of 3D corrections that included model atmospheres accounting for the presence of hydrogen. Their findings showed that the 3D \log\textsl{{g}} corrections were most substantial for T_{\rm eff}\lesssim 18,000 K, reaching up to -0.2 dex at \log\textsl{{g}}=8.0 . However, since \log\textsl{{g}} determinations in this temperature range could also be significantly influenced by the treatment of broadening by neutral particles, as well as non-ideal effects from helium perturbation by neutral atoms, it remains challenging at this stage to draw firm conclusions regarding the validity of these 3D corrections. Further improvements in understanding the microphysics of helium are needed to reduce these uncertainties. In this spirit, we decided to undertake an investigation by reevaluating the physics of line broadening theory. As a first step, we began by refining the Stark broadening calculations for neutral helium lines by 5 , which are the most widely used tables in the spectroscopic analyses of DB white dwarfs. These calculations rely on the standard Stark broadening theory with the inclusion of the one-electron theory in the far wings (see below). 56 introduced new calculations of Stark-broadened profiles for neutral helium lines using computer simulations that unify the treatment of ions and electrons. These simulations also incorporate corrections for ion dynamics, transitions of electron contributions from the core to the wings of the profile, numerical integration of the time evolution operator for helium perturbed by a fluctuating electric field, the Debye correction for charge-motion correlations, local density variations, and particle reinjection. 57 have applied a similar, but more advanced, framework to generate a comprehensive set of Stark profile calculations for most neutral helium lines relevant to the analysis of DB white dwarf optical spectra in the 3800 to 5200 Å wavelength range. In this paper, we present a comprehensive approach to understanding the effects of line broadening on the spectroscopic analysis of DB white dwarfs. In Section II , we summarize the key physical components of line broadening theory for neutral helium, with particular emphasis on the Stark broadening of He i line profiles. Section III describes the theoretical framework and observational data used to test the impact of various elements within the line broadening theory, which are then individually examined in Section IV . Our conclusions follow in Section V . II Calculation of He i Line Profiles The spectroscopic technique used for DB stars typically relies on optical spectra, where \rm{H}{\alpha} , if present, helps to constrain the hydrogen abundance, while the blue portion of the spectrum is used to determine T_{\rm eff} and \log\textsl{{g}} (see, e.g., 13 ). In this section, we briefly discuss some of the physical components involved in calculating model spectra for determining the atmospheric parameters of DB white dwarfs. The spectra of DB white dwarfs are primarily dominated by neutral helium lines, with occasional traces of hydrogen, mostly as \rm{H}{\alpha} . The main broadening mechanism for He i lines in these stars is Stark broadening, caused by charged perturbers in the surrounding plasma. The most widely used calculations for Stark-broadened profiles were done by B97 over 25 years ago, and these have remained the standard in the field. These profiles are computed using the standard Stark broadening theory as implemented by 3 and 4 , with the inclusion of the one-electron theory of 2 ; see also 56 and Paper I for a review of the theory and a discussion on these various approximations. Stark profiles for 21 He i lines are provided as tables covering a range of electronic densities ( N_{e}=10^{14} to 6\times 10^{17} cm 3 ), temperatures (10,000 K, 20,000 K, and 40,000 K), and grids of 40 to 70 frequency points. Convolution with Doppler broadening is also included. Another crucial aspect of these calculations is the inclusion of line dissolution. In brief, rather than extending the integral over the ionic field to infinity, it is carried out only up to the critical value at which the highest Stark levels for a given n merge with other levels (see equations 4.1 and 5.5 of B97 following the method originally proposed by 52 . In this approach, the critical electric field is calculated using the theory introduced in the occupation probability formalism by 40 . Note that a similar method was used by 53 in their Stark broadening calculations for hydrogen lines. As discussed above, 56 presented new calculations of Stark-broadened profiles for He i lines using computer simulations that eliminated several approximations found in the earlier work of B97, with the most significant improvement being the inclusion of ion dynamics. However, line dissolution has not yet been incorporated into these calculations. We evaluate these effects separately in Section IV . Below T_{\rm eff}\sim 16,000 K, He i line broadening by neutral particles begins to dominate over Stark broadening. The implementation of van der Waals and resonance broadening in our model calculations is described at length in 6 and summarized in Section 3.2 of 29 . Unfortunately, broadening by neutral particles introduces significant uncertainty in estimating white dwarf masses using the spectroscopic technique. This is illustrated in Figure 11 of 29 , which compares the spectroscopic mass distributions of DB white dwarfs in the SDSS across effective temperatures, using the van der Waals broadening theory of 58 and 22 . It is challenging to determine which theory is more suitable until other issues, including the impact of 3D hydrodynamical corrections on spectroscopic \log\textsl{{g}} determinations, are fully addressed. Complicating matters further, at such low temperatures, He i lines eventually disappear, rendering the spectroscopic technique unreliable. At this point, the photometric technique should be used instead. III Spectroscopic and Photometric Analysis The spectroscopic and photometric data for our analysis are taken from the Data Release 17 (DR17) of SDSS. We selected all DB spectra with a signal-to-noise ratio above {\rm S/N}=20 , resulting in a total of 738 objects. All magnetic white dwarfs were excluded from our analysis. For the photometric analysis, we combined ugriz photometry with Gaia DR3 parallax measurements. Our model atmospheres are detailed in 13 , with updates to broadening by neutral particles discussed in 29 . The model spectra are calculated using various Stark-broadened profiles for He i lines, which are described in the following section. Note that in our comparative analysis, the temperature and pressure structures remain identical across all cases, as the specifics of the line profile calculations have no significant effect on the atmospheric structure — something we verified retrospectively. Our aim in this paper is to explore various approximations within the He i broadening theory. To achieve this, we will compare the spectroscopic results, T_{\rm eff} and \log\textsl{{g}} , obtained under different approximations to better understand the systematic effects introduced by each physical component. Additionally, we will examine the global properties of the DB sample by comparing photometric and spectroscopic mass distributions as a function of effective temperature. Our determinations of the photometric and spectroscopic parameters are illustrated in Figure 1 . In the top panel, the ugriz photometry, converted into absolute fluxes using the AB calibration, is compared to model spectra predictions, averaged over each filter bandpass. Interstellar reddening is accounted for using STILISM values ( 43 ; 15 ) . The fitted parameters here are the effective temperature, T_{\rm eff} , and the solid angle, \pi(R/D)^{2} , where R is the star’s radius and D is its distance from Earth, obtained from the Gaia parallax measurement. The radius is then converted into stellar mass using the cooling sequences from 7 with C/O-core (50/50) compositions, q({\rm He})\equiv M_{\rm He}/M_{\star}=10^{-2} , and q({\rm H})=10^{-10} , appropriate for non-DA stars. We first fit the spectral energy distribution assuming a pure helium composition. The resulting photometric temperature is then used as an initial estimate to fit the SDSS spectrum, thereby avoiding the issue of choosing between the typically ambiguous cool and hot solutions that arise when using only the spectroscopic technique — one on either side of the peak strength of the neutral helium lines, which occurs near 25,000 K for DB white dwarfs (see Section 4.2 of 13 for further discussion). The measured hydrogen-to-helium abundance ratio (H/He by number) is then used to obtain updated estimates of the photometric parameters. Figure 1: Best photometric and spectroscopic fits to a typical DB white dwarf in the DR17 SDSS sample. In the top panel, the observed ugriz photometry is shown as error bars (1 \sigma uncertainty), while the model atmosphere fit is shown as solid dots, with the parameters given in the panel. The H/He abundance ratio in this fit is constrained from the spectroscopic fit below. The lower panel shows our spectroscopic fit, with the observed and model spectra displayed in black and red, respectively, both normalized to a continuum set to unity. The spectroscopic fit to the H \alpha line profile, shown in the insert, is used to measure, or constrain, the hydrogen abundance of the overall solution. The spectroscopic technique applied to DB/DBA white dwarfs follows the method described in detail by 13 and relies on normalized spectra, as illustrated in the bottom panel of Figure 1 . Briefly, the normalization procedure for DB spectra utilizes our synthetic spectra and a high-order polynomial (up to \lambda^{5} ) to produce a smooth fitting function for determining the continuum location. Unlike the approach in Bergeron et al., here we use the photometric temperature to determine on which side of the maximum line strength the solution lies. Once the observed spectrum is normalized, we obtain an initial estimate of T_{\rm eff} and \log\textsl{{g}} using the blue portion of the spectrum ( 3800-5200 Å). Keeping these atmospheric parameters fixed, we then determine (or constrain) the hydrogen abundance by fitting the region near H \alpha ( 6400-6800 Å). This process is repeated until the {\rm H}/{\rm He} value converges. When H \alpha is undetectable, only upper limits on the hydrogen abundance can be obtained. The \log\textsl{{g}} value is then converted into stellar mass using the same cooling sequences as before. As noted above, the spectroscopically determined H/He value is also used for the photometric fit displayed in the top panel of Figure 1 . With these determinations, we can now make various comparisons between different He i profiles using the stellar parameters derived from both the spectroscopic and photometric methods. IV Results IV.1 Reappraisal of the Semi-Analytical He i Line Profiles Since the publication of our Stark profiles in B97, we have continuously maintained and improved our code, although no new calculations have been officially released. Specifically, the code has been rewritten from Fortran to C++, and minor errors have been identified and corrected. Calculation precision has been enhanced by redefining a small number of constants, originally specified as single-precision, to double-precision. Additionally, the convergence criteria for numerical integrals (over ionic microfields and for convolution with a Doppler profile) have been tightened, and the interpolation procedure for Hooper tables ( 39 ) , which describe the microfield distribution as a function of thermodynamic conditions, has been improved. Two new options have been added to the code. The effect of line dissolution can now be included (or excluded) in the calculation of profiles by controlling the upper limit of the integral over the ionic microfield. Moreover, the electron contribution to broadening can either be described by impact broadening for the full profile or switch to the one-electron regime in the wings. The originally published profiles included line dissolution and combined both electronic regimes. To ensure direct comparison between the semi-analytical and simulation methods presented in this article, the mixing of the upper transition levels due to interactions with the electric field of the perturbers (ions and electrons) is the same for both sets of profiles, and has been restricted to levels with the same principal quantum number n as the upper level of the allowed transition. We first present a comparison of the results obtained between both sets of calculations. Figure 2 compares the T_{\rm eff} and \log\textsl{{g}} values for the DB white dwarfs in the DR17 SDSS sample obtained from spectroscopic fits using the Stark profile calculations from B97 alongside our updated calculations described above. Both sets of calculations consider line dissolution, Doppler broadening, resonance and van der Waals broadening, following a modified version of the theory by 22 as described in Section IV.6 . Additionally, both profile sets rely on tables with identical frequency sampling. Note that in Figure 2 (and subsequent figures), we have restricted the range of T_{\rm eff} and \log\textsl{{g}} values to the field of interest, although objects exist outside this range, particularly at higher effective temperatures. Figure 2: Comparison of effective temperatures and surface gravities for the DB white dwarfs in the DR17 SDSS sample obtained from spectroscopic fits using the Stark profile calculations from B97 and our revised calculations, as described in the text. The dotted lines indicate the 1:1 correspondence. The comparison shown in Figure 2 indicates that our revised calculations are nearly identical to those of B97, with the updated T_{\rm eff} values being, on average, only 2.5 K lower and the updated \log\textsl{{g}} values just 0.007 higher. Figure 3 further presents the spectroscopic mass distribution as a function of T_{\rm eff} for the same sample, comparing it with the photometric mass distribution. As a reference, we show the canonical mean mass of 0.6 M_{\odot} as a dashed line in each panel and a constant mass of 0.55 M_{\odot} as a dotted line. The latter corresponds to the empirical mass threshold below which common-envelope evolution must be invoked to explain the presence of low-mass white dwarfs, based on the photometric mass distribution of over 4000 white dwarfs analyzed by 42 . Consequently, white dwarfs evolving through a single-star channel should lie above this mass limit, as is the case for most DB white dwarfs observed in the photometric mass distribution (upper panel of Figure 3 ). The results displayed in Figure 3 are entirely consistent with those presented in Figures 8 and 11 of 29 and, more importantly, Figure 7 of 30 . We briefly summarize the conclusions of these studies here, and direct the reader to Section 5.1 of 30 for a more detailed discussion. While the photometric mass distribution remains well-centered on the canonical mean mass of approximately 0.6 M_{\odot} for white dwarfs across the full T_{\rm eff} range shown in Figure 3 , the spectroscopic mass distribution deviates significantly from the 0.6 M_{\odot} value in several temperature ranges. For T_{\rm eff}\gtrsim 27,000 K, the low spectroscopic masses are thought to be artifacts due to calibration issues with the SDSS spectra, a problem also noted by 55 and 28 in the context of DA white dwarfs. In the range 22,000~{\rm K}>T_{\rm eff}>16,000 K, the spectroscopic masses appear overestimated relative to the photometric values; this corresponds to a temperature range where 3D hydrodynamical corrections are expected to play an important role, as suggested by 19 ; 20 . Note that these 3D corrections have not been incorporated into our results and will be addressed below. Finally, for T_{\rm eff}\lesssim 14,000 K, the spectroscopic distribution shows a notable increase in both mass and dispersion, especially when compared to the photometric mass distribution. This high-mass or high- \log\textsl{{g}} issue is typically attributed to limitations in the broadening theory by neutral particles, which we discuss further below. Figure 3: Photometric (upper panel) and spectroscopic (bottom panel) mass distributions as a function of T_{\rm eff} for the DB white dwarfs in the DR17 SDSS sample with {\rm S/N}>20 . The spectroscopic parameters have been obtained using model spectra that include our revised Stark broadening profiles (see text and Figure 2 ) based on the approach described in B97. The objects with an equivalent width {\rm EW}<5 Å for He i \lambda 4471 are shown by red symbols. In both panels, the dashed and dotted lines correspond to a constant mass of 0.6 and 0.55 M_{\odot} , respectively. We reexamined this high-mass issue using our analysis results and a detailed inspection of our spectroscopic fits. It became evident that the high-mass white dwarfs in Figure 3 correspond to objects with the weakest helium lines. Similar to DA stars, below a certain T_{\rm eff} threshold — also dependent on \log\textsl{{g}} — the absorption lines weaken to the point where the spectroscopic technique becomes unreliable. To quantify this effect for DB stars, we measured the equivalent width (EW) around He i \lambda 4471 (from 4220 to 4625 Å) for all objects in our sample. The DB stars with {\rm EW}<5 Å are highlighted in red in Figure 3 . Below this threshold, He i \lambda 4471 becomes one of the few lines detectable in the blue portion of the spectrum (as used in Figure 1 ). Excluding these objects from our analysis reveals in Figure 3 that the high-mass issue almost completely disappears. This conclusion, of course, likely depends on the restricted {\rm S/N}>20 range of our spectroscopic sample. From this point forward, the B97 profiles will refer to the revised calculations described in this section. IV.2 Frequency Sampling and Doppler Broadening As explained in the Introduction, He i Stark profiles are typically provided in the form of tables across a range of electron densities, temperatures, and a grid of frequency (or wavelength) points. Our original tables contained 40 to 70 frequency points, depending on each He i line. In the course of our spectroscopic analysis, we discovered that the precise number of frequency points in these tables was particularly important, especially for forbidden transitions and very narrow absorption lines such as He i \lambda 3889, which is notably \log\textsl{{g}} -sensitive in cool DB white dwarfs. More importantly, we also identified significant issues with the convolution of our profiles with Doppler broadening, which we now discuss in detail. Tables of Stark profiles are commonly provided both with and without convolution with a Doppler profile (see, e.g., the case of hydrogen lines by 59 ), which accounts for the atom’s velocity along the line of sight ( 48 ) . The tables of B97 were only available to the scientific community with Doppler broadening included. In our recent experiments using revised calculations with the original (1X) frequency sampling from the tables of B97 — displayed in Figure 2 — and double (2X) the frequency sampling, we found that our Doppler profile convolution was problematic, particularly for extremely narrow lines, such as the \log\textsl{{g}} -sensitive He i \lambda 3889. The problem, along with the solution, is illustrated in Figure 4 , where we show various intensity profiles for He i \lambda 3889 at T=20{,}000 K and N_{e}=10^{14} cm -3 . The blue line in Figure 4 shows the unconvolved profile sampled on an extremely fine wavelength grid, indicated by small blue dots (note that the wavelength range in this figure is approximately 1 Å wide). The convolution of this profile with a Doppler profile is shown as the red line, with the same wavelength sampling marked by red dots. As explained above and in Paper I, we now favor twice the frequency sampling (2X) used in B97 to better represent the core of some allowed and forbidden components of He i , especially at low densities. The resampling of the convolved profile on this coarser grid is shown as a black dashed line, with open circles indicating the wavelength sampling. The agreement with the convolved profile on the fine wavelength grid (red line) is excellent, demonstrating that the overall procedure is sufficiently accurate, and this is the approach we now adopt. Figure 4: Comparison of various Stark intensity profiles for He i \lambda 3889 at T=20,000 K and N_{e}=10^{14} cm -3 . The blue line corresponds to the unconvolved profile sampled on an extremely fine wavelength grid marked by small blue dots. The red line (with small dots) is the same profile convolved with a Doppler profile, while the black dashed line corresponds to the resampling of this convolved profile on our 2X wavelength grid (see text) marked by open circles. The magenta line corresponds to the unconvolved profile sampled on the 2X wavelength grid marked by open circles, while the black solid line corresponds to the direct convolution of this profile with a Doppler profile (using the same wavelength grid). Finally, the green line shows the convolved line profile using the original 1X wavelength grid from B97 marked by open circles. The issue with the earlier calculations of B97 is illustrated by the magenta profile in Figure 4 , which shows the unconvolved Stark profile sampled on the coarse 2X wavelength grid, represented by open magenta circles. At this particular low electron density, this undersampled profile is a poor representation of the true profile shown in blue. As a direct consequence of this undersampled profile, the resulting convolution with a Doppler profile — shown by the black solid line and black open circles — yields a higher intensity near the line core (and consequently a deeper absorption line) when compared to our more accurate calculations (dashed black line). The situation is even worse with the convolved profiles using the original (1X) sampling of B97, shown by the green line and open green circles; here, a crucial point is even missing at \Delta\lambda=0 , which further exacerbates the problem. We now provide a detailed comparison of the atmospheric parameters of DB stars in the DR17 SDSS sample using the various calculations described in Figure 4 . In Figure 5 , we compare the \log\textsl{{g}} values derived from synthetic spectra using the B97 version of our calculations with corrected Doppler broadening (labeled new Doppler 2X), as described above, and the earlier B97 version with inaccurate Doppler broadening (labeled old Doppler) — described in Section IV.1 and used in Figure 2 — with the original (1X, bottom panel) and double (2X, top panel) the frequency sampling used in the tables of B97; the effects on T_{\rm eff} are negligible and are therefore not shown. The comparisons indicate that \log\textsl{{g}} values differ most significantly for cool DB white dwarfs with T_{\rm eff}<15,000 K, shown in red in both panels, a direct consequence of our corrected treatment of Doppler broadening. The results with double (2X) the frequency sampling (top panel) show a significant scatter in this range of temperature, but with only a small systematic shift in \log\textsl{{g}} . However, the effects become more pronounced when we compare the results with the original (1X) frequency sampling (bottom panel). In this case, our corrected treatment of Doppler broadening predicts \log\textsl{{g}} values that are, on average, approximately 0.1 dex lower for cool DB white dwarfs. Figure 5: Comparison of surface gravities for the DB white dwarfs in the DR17 SDSS sample obtained from spectroscopic fits using our revised version of our semi-analytical Stark profile calculations where Doppler broadening is included accurately (new Doppler 2X; see text) and from our previous version (old Doppler) with 1X (bottom panel) and 2X (top panel) the original frequency sampling of B97. Objects with T_{\rm eff}<15,000 K are shown in red, while the dotted line indicates the 1:1 correspondence. We can gain more insight by comparing our best spectroscopic fits to a specific cool DB white dwarf using the various calculations described in Figure 4 . We show such a fit in Figure 6 using the calculations with the proper inclusion of Doppler broadening and frequency sampling (2X, blue line), and the improper inclusion of Doppler broadening with 1X (green line) and 2X (red line) the original frequency sampling of B97. The most significant differences between these fits occur for He i \lambda 3889 and \lambda 6678. As seen for He i \lambda 3889, our corrected calculations accurately predict the depth of the absorption profile, whereas the 1X and 2X profiles are predicted to be too shallow and too deep, respectively, consistent with the intensity profiles shown in Figure 4 . A similar conclusion applies to He i \lambda 6678. For this particular object, the \log\textsl{{g}} value obtained from our corrected profiles is the same as that obtained with the 2X grid. However, the surface gravity measured using the 1X grid is 0.17 dex higher, in line with the systematic shift observed in the bottom panel of Figure 5 . The reason for this difference is that synthetic spectra based on the 1X frequency sampling (green profile in Figure 4 ) are simply unable to reproduce the observed deep line core, even by lowering the \log\textsl{{g}} value; the solution is thus driven by other lines in the spectrum. Figure 6: Best spectroscopic fits to the DB white dwarf SDSS J161301.55+234830.7 with model spectra calculated with the B97 version of the Stark profiles with corrected Doppler broadening (bottom, blue) and the B97 version with the old (and inaccurate) Doppler broadening version with 2X (middle, red) and 1X (top, green) the frequency sampling of the original tables from B97. The values of T_{\rm eff} , \log\textsl{{g}} , and \log{\rm H}/{\rm He} of each fit are given in the figure. The insert shows the corresponding region around He i \lambda 6678. To summarize this subsection, we adopt in what follows improved semi-analytical He i Stark profiles calculated using twice (2X) the frequency sampling of the original grid of B97, including additional frequencies tightly packed around the line core (as shown by the blue dots in Figure 4 ). This grid is then used to calculate the convolution with a Doppler profile (red dots in Figure 4 ), and for the final grid, resampled on the 2X frequency grid (the black dashed line and open circles in Figure 4 ). This has been done for all lines, and not just for He i \lambda 3889. IV.3 Line Profile Normalization The semi-analytical Stark profiles from B97 include quasi-static broadening by ions together with electron broadening. The latter is treated using the impact approximation near the line core and the one-electron approximation in the far wings, where strong collisions play a more significant role. An interpolation scheme is applied to ensure a smooth transition between the one-electron and impact regimes. As discussed extensively in Paper I, while the Stark profiles derived solely from the impact theory are, by definition, properly normalized, this is not the case when the one-electron approximation is used. This lack of normalization was identified in Paper I as one of the main sources of discrepancy between our semi-analytical profiles and those obtained from our computer simulations, in particular at high density. This finding eventually led us to improve our semi-analytical profiles by implementing a detailed normalization procedure (see Section 3.3 of Paper I). We illustrate the impact of this normalization procedure in Figure 7 , which compares the effective temperatures and surface gravities for the DB white dwarfs in the DR17 SDSS sample obtained using spectra calculated from the unnormalized grid of semi-analytical He i Stark profiles (as described above and used previously) with those derived from the same profiles after proper normalization. While the effects on the effective temperatures are negligible, the \log\textsl{{g}} values obtained from the normalized profiles are, on average, about 0.03 dex lower, with the largest differences observed for white dwarfs hotter than 15,000 K. Figure 7: Comparison of effective temperatures and surface gravities for the DB white dwarfs in the DR17 SDSS sample obtained from spectroscopic fits using the revised version of our semi-analytical Stark profile calculations with and without normalization (see text). Objects with T_{\rm eff}<15,000 K are shown in red, while the dotted line indicates the 1:1 correspondence. This normalization procedure represents the final improvement implemented in the semi-analytical approach originally developed by B97. We designate these improved Stark profiles as B25, which will serve as our reference framework for the comparisons presented in the remainder of this paper. IV.4 Line Dissolution As briefly discussed in the Introduction, another important aspect of line broadening calculations is the inclusion of line dissolution, originally introduced by 52 . This concept was first applied in the context of He i Stark profiles by B97. In summary, the most general form of the profile, using the quasi-static approximation for ions and the impact approximation for electrons, is the following average over the ion microfield distribution of microfield-dependent electron-broadened profiles ( 37 ; 3 ; 4 ) , I(\omega)=\int_{0}^{\infty}dF\ W_{a}(F)\ I_{e}(\omega,F)\ , (1) where I_{e}(\omega,F) is the electron-broadened profile at frequency \omega , W_{a}(F) is the microfield distribution parameterized by a — the ratio of the mean distance between perturbers to the Debye radius —, and F is the electric field. To account for the effects of line dissolution, the upper boundary in this integral is replaced by F^{\rm crit} , the critical field value at which the highest Stark levels with a given n merge with other levels. This critical field value is obtained for the excited levels of the He i atom using the method of 52 , based on the theory of 40 in the context of their occupation probability formalism (see B97 for further details). This treatment does not account for the additionnal dissolution induced by interactions with neutral particles. Incorporating this supplementary contribution would require substantial modifications to the semi‑analytical theory, as this effect cannot be captured through a simple truncation of the ionic electric field. To our knowledge, no studies addressing this issue are currently available. As mentioned previously, a similar approach was used by 53 in their Stark broadening calculations for hydrogen lines (see their Section 3). Note, however, that our grid of model spectra are calculated using the occupation formalism of 40 for both the hydrogen and helium populations and corresponding bound–bound, bound–free, and pseudo-continuum opacities, as implemented by 21 , and that in that context, the occupation probabilities include the contributions of both charged and neutral particles (see also Section IV of 14 ). To study the effects of line dissolution on the He i Stark profiles, we calculated a table of line profiles similar to B25 but by excluding this particular process. The spectroscopic \log\textsl{{g}} values obtained with and without line dissolution are compared in the top panel of Figure 8 , while the bottom panels show the corresponding mass distributions as a function of T_{\rm eff} ; the spectroscopic T_{\rm eff} values are not significantly affected and are therefore not compared here. Our results indicate that neglecting line dissolution reduces the \log\textsl{{g}} values by only about 0.05 dex (or \sim 0.03 M_{\odot} in mass), and this effect is limited to hot ( T_{\rm eff}>15,000 K) DB white dwarfs. Although the effect of line dissolution is not negligible, it remains small compared to other sources of uncertainty. Figure 8: Top panel: Comparison of surface gravities for the DB white dwarfs in the DR17 SDSS sample obtained from spectroscopic fits using our updated Stark profiles (B25) with and without line dissolution taken into account in the calculations; objects with T_{\rm eff}<15,000 K are shown in red, while the dotted line indicates the 1:1 correspondence. Bottom panels: Spectroscopic mass distributions as a function of T_{\rm eff} for the same sample using He i profiles with and without line dissolution; the dashed and dotted lines correspond to a constant mass of 0.6 and 0.55 M_{\odot} , respectively. IV.5 He i Line Profiles from Computer Simulations In the semi-analytical approach of B97 for calculating the Stark profiles of neutral helium, electrons were treated as dynamic and ions as static. With advancements in computational power, other research groups have significantly enhanced the treatment of Stark broadening by conducting computer simulations that detail the dynamics and interactions of the perturbers (ions and electrons) near the emitter — see in particular 33 , 32 , and 35 in the case of the hydrogen atom. Stark profiles for the helium atom, however, were generated for only two lines, He i \lambda 4471 and \lambda 4922 ( 34 ; 44 ) , limiting their applicability to the computation of synthetic spectra for white dwarf stars. In 56 , we introduced our own simulation environment for calculating He i Stark profiles, incorporating several key aspects from previous work to better represent the dynamical environment of the helium atom. These aspects include the unification of ion and electron treatments, corrections for ion dynamics, the proper treatment of electron broadening from the core to the wing, still assuming the semi-classical approximation, the numerical integration of the quantum operator governing the time evolution of helium perturbed by a fluctuating electric field, the Debye correction for the correlation of charged perturbers’ motions, as well as local density variations and particle reinjection. Intermediate steps, such as constructing the simulation space and the quasi-static model, allowed us to validate the simulation space and ensure the system’s statistical consistency. In that exploratory work, we generated grids of Stark profiles for the two most important optical lines of the helium atom, namely He i \lambda 4471 and \lambda 4922, for temperatures between 10,000 K and 40,000 K and electron densities ranging from 1\times 10^{14}\,\mathrm{cm}^{-3} to 6\times 10^{17}\,\mathrm{cm}^{-3} . More recently, in Paper I, we improved these simulations and produced an extensive grid of Stark profiles for 12 neutral helium lines covering the same range of temperatures and electron densities, and a 13th line (He i \lambda 4713) with the same temperatures but with electron densities from 3\times 10^{15}\,\mathrm{cm}^{-3} to 6\times 10^{17}\,\mathrm{cm}^{-3} due to its narrow nature, as explained in detail in Paper I. The key improvement in these calculations was the application of the power spectrum equation for line profile generation, rather than the autocorrelation function used in the earlier calculations of 56 . This helped to reduce the noise level in our calculations, especially in the far wings, as confirmed mathematically by 51 and demonstrated by 18 for hydrogen lines, thus enhancing the time-quality ratio of the method. This advancement enabled, for the first time, more realistic calculations of synthetic spectra for white dwarf stars. However, the process of line dissolution, as described in Section IV.4 , was not included in these line profile calculations. We compare in Figure 9 the effective temperatures and surface gravities of the DB white dwarfs in the DR17 SDSS sample, using He i profiles based on our new computer simulations and the B25 Stark profiles. The Deridder & van Rensbergen formalism for van der Waals broadening (see Section IV.6 ) is applied in both sets of calculations. Given our simulation-based profiles do not account for line dissolution, we also exclude this process in the B25 calculations (labeled as NLD in the following figures, for “no line dissolution”). Our results show that the effective temperatures are nearly identical, with an average difference of less than 130 K. The \log\textsl{{g}} values, however, display a systematic offset, with the Stark profiles derived from computer simulations yielding surface gravities about 0.03 dex higher than those obtained from the semi-analytical profiles, regardless of T_{\rm eff} . Figure 9: Comparison of effective temperatures and surface gravities for the DB white dwarfs in the DR17 SDSS sample obtained from spectroscopic fits using the B25 Stark profiles with no line dissolution (NLD) and the computer simulations from Paper I. Objects with T_{\rm eff}<15,000 K are shown in red, while the dotted line indicates the 1:1 correspondence. Figure 10: Bottom panel: Comparison of normalized synthetic spectra (Eddington fluxes) at T_{\rm eff}=20,000 K and pure helium compositions. The top two spectra show the calculations at \log\textsl{{g}}=8.0 obtained with the B25 Stark profiles with no line dissolution (black) and the computer simulations from Paper I (red), while the bottom spectra (shifted arbitrarily by 0.4) also rely on the Stark profiles from computer simulations but for different \log\textsl{{g}} values indicated in the figure. All spectra are convolved with a 3 Å (FWHM) Gaussian profile. Top panel: Residual fluxes of the \log\textsl{{g}}=8 normalized spectra. In Figure 18 of Paper I, we compared the absolute Eddington fluxes of white dwarf model spectra at various effective temperatures and \log\textsl{{g}}=8 , using both sets of He i Stark profiles. Because the spectroscopic method relies on normalized spectra rather than absolute fluxes, we compare instead in Figure 10 pure-helium DB model spectra at T_{\rm eff}=20,000 K, but normalized to a continuum set to unity. The comparison is restricted to the wavelength range where the He i lines are particularly sensitive to \log\textsl{{g}} . The top two spectra in the bottom panel contrast the B25 Stark profiles at \log\textsl{{g}}=8.0 with no line dissolution (black) and the computer simulations from Paper I (red). Given these overlap almost perfectly, we also show in the top panel the residual normalized fluxes between these two spectra. The small differences, of order 1%, are likely related to the treatment of ion dynamics near the line cores and to electron broadening between the line cores and the continuum, which is handled in a significantly more sophisticated manner in the simulation-based approach. Note that there are no differences in the continuum regions because this is where the spectra are normalized. Also shown in the bottom panel of Figure 10 are model spectra computed with Stark profiles from computer simulations but for different \log\textsl{{g}} values. These illustrate how subtle differences in the overall line profiles can affect the derived spectroscopic \log\textsl{{g}} values. We further compare in Figure 11 the spectroscopic mass distributions obtained from He i Stark profiles based on our computer simulations and the B25 semi-analytical calculations for the DB white dwarfs in the SDSS sample. Also shown in red are updated results for the DB white dwarfs analyzed by 13 and 50 , based on single-slit spectroscopic data (see next paragraph). The differences between both sets of calculations are barely noticeable in these diagrams, as anticipated from the results shown in Figure 9 . We still observe, however, that the spectroscopic masses in the \sim 18,000–21,000 K temperature range remain overestimated compared with the canonical average mass of 0.6 M_{\odot} for white dwarfs. At lower temperatures, the masses are now underestimated when using the van der Waals theory of Deridder & van Rensbergen. These masses can be increased, however, if the Unsöld theory is adopted instead, an alternative approach explored in the next subsection. Figure 11: Spectroscopic mass distributions as a function of T_{\rm eff} for the DB white dwarfs in the DR17 SDSS sample (white circles) using the B25 Stark profiles and the computer simulations from Paper I. In both sets of calculations, line dissolution is omitted (NLD), and van der Waals broadening is included using the Deridder & van Rensbergen theory. Also shown in red are the results for the DB white dwarfs analyzed by 13 and 50 . In both panels, the dashed and dotted lines correspond to constant masses of 0.6 and 0.55 M_{\odot} , respectively. Finally, as mentioned in the Introduction, residual flux calibration issues with the SDSS spectra have often been invoked to explain some of the discrepancies in the measured spectroscopic parameters of both DA and DB white dwarfs (see, e.g., 28 ; 30 ). However, as shown in Figure 11 , the results obtained for the DB white dwarfs analyzed by 13 and 50 , based on single-slit spectroscopic data, now agree perfectly with the SDSS results. This strongly suggests that the flux calibration problem affecting earlier SDSS spectra has now been corrected. Consequently, flux calibration issues can no longer be invoked to explain the discrepancies in the spectroscopic mass distribution of DB white dwarfs. IV.6 Broadening by Neutral Particles and 3D Hydrodynamical Corrections Given that the Stark profiles from B97 have been widely used in the spectroscopic analysis of DB white dwarfs, some of the results based on these earlier calculations need to be revisited, especially for cool DB stars. As discussed above, we now rely exclusively on the B25 profile calculations with the proper inclusion of Doppler broadening and normalization, with twice the frequency sampling (2X) of the original tables of B97. In this section, we now attempt to bring all our results together in terms of He i Stark profiles, broadening by neutral particles, and 3D hydrodynamical corrections. For completeness, we first begin by summarizing the two different He i line broadening theories by neutral particles used in our analysis. In our models, we include only broadening by ground-state He i perturbers, which dominate cool DB atmospheres. Resonance broadening arises from a first-order resonant dipole-dipole interaction V(r)\propto r^{-3} between identical atoms, and requires a dipole-allowed transition between the ground state and either level of the radiative transition. Van der Waals broadening results from second-order induced dipole-dipole interactions, and is always present. These definitions yield three cases for He i line broadening by neutral perturbers: (1) For different atomic species, only van der Waals broadening applies and its contribution from elements other than helium is negligible in DB/DBA stars; (2) For He i lines connecting triplet states (e.g., \lambda 4471), no dipole-allowed transition connects the ground state to either level of the transition, so only van der Waals broadening occurs; (3) For singlet He i lines considered in this work (e.g., \lambda 4922), resonance broadening is present because the lower or upper level connects to the ground state, and the full interaction includes both first- and second-order terms. In our models, van der Waals and resonance broadening are combined following the procedure described in detail by 6 . Resonance broadening is treated according to the theory of 1 , while van der Waals broadening is taken into account using the theory from 58 , or an alternative approach we refer to as the Deridder & van Rensbergen formalism. First, the width of the Lorentzian profile ( \omega_{\rm vdW} ) is calculated twice: once using the theory in 58 and again by applying the approach described in 22 . The Deridder & van Rensbergen theory consistently predicts a broader profile when the initial and final effective quantum numbers of the transition exceed 2 or 3. Because the Smirnov potential, which they used, is no longer valid below these quantum numbers, we retain the following conservative value for the width of the Lorentzian profile: \omega_{\rm vdW}={\rm max}(\omega_{\rm Unsold},\omega_{\rm Deridder})\ . (2) Empirically, 6 found that the neutral helium lines in cool DB stars at \lambda=4121 Å and 4713 Å could be more accurately reproduced when treated strictly within the 58 theory. This is the approach adopted here as well. Regarding transitions between singlet states, the available data in the literature, to our knowledge, only provide separate estimates for resonance and van der Waals broadening contributions. This raises the issue of statistical dependence: it is not guaranteed that these two mechanisms are independent, since a given collision may contribute to both. Furthermore, assuming strong statistical dependance, it is not clear whether the two contributions should add constructively or destructively. Rather than summing the two contributions to compute the Lorentzian half-width—which would implicitly assume statistical independence—we chose to retain the maximum of the two half-widths, \omega_{\rm neutral}={\rm max}(\omega_{\rm resonance},\omega_{\rm vdW}) (3) rather than following the prescription by 45 , who determined that combining resonance and van der Waals broadening resulted in a profile with \omega_{\rm neutral}\sim 0.6-0.8\,(\omega_{\rm resonance}+\omega_{\rm vdW}) . However, that study was limited to temperatures around 100 K, which are not applicable to our DB models. We first show in Figure 12 the spectroscopic mass distributions as a function of T_{\rm eff} for our DB white dwarf SDSS sample using the B25 Stark profiles, which, for the purpose of this discussion, now include line dissolution. These calculations thus represent the state-of-the-art Stark profiles within the semi-analytical framework of B97. The top two panels rely on the different treatments of van der Waals broadening described above, namely the 58 theory and the Deridder & van Rensbergen formalism (middle panel), while the bottom panel also adopts the Unsöld theory but includes the 3D hydrodynamical corrections of 20 . Figure 12: Spectroscopic mass distributions as a function of T_{\rm eff} for the DB white dwarfs in the DR17 SDSS sample using the B25 Stark profiles, including line dissolution. The top and middle panels differ in terms of the theory used for van der Waals broadening, as indicated in each panel, while the bottom panel is similar to the top panel but with the 3D hydrodynamical corrections of 20 taken into account. In all panels, the dashed and dotted lines correspond to constant masses of 0.6 and 0.55 M_{\odot} , respectively. The effect of our corrected treatment of Doppler broadening on the mass distribution can be appreciated by comparing the middle panel of Figure 12 with the bottom panel of Figure 3 , which relies on the same van der Waals broadening theory but differs in terms of the way Doppler broadening is taken into acount; note that for clarity, DB stars with weak He i lines ( {\rm EW}_{4471}<5 Å) are not shown in Figure 12 . While our earlier semi-analytical profiles with inaccurate Doppler broadening favored the Deridder & van Rensbergen treatment of van der Waals broadening, our revised B25 profiles used in the middle panel of Figure 12 clearly show that this approach now yields masses that are too low below T_{\rm eff}\sim 17,000 K. The Unsöld theory (top panel), on the other hand, produces masses that appear slightly overestimated. Given that 3D hydrodynamical simulations tend to lower the inferred \log\textsl{{g}} values compared to those derived from 1D model atmospheres (see Figure 5 of 20 ), we can significantly reduce the spectroscopic masses by applying the 3D hydrodynamical corrections to the Unsöld determinations, as shown in the bottom panel of Figure 12 . The resulting mass distribution below T_{\rm eff}\sim 17,000 K now follows the canonical 0.6 M_{\odot} mean mass much more closely — in better agreement with the photometric mass distribution displayed in the top panel of Figure 3 . However, a clear discrepancy remains in the range 24,000~{\rm K}>T_{\rm eff}>17,000 K, where the masses are still predicted to be too high, even after including the 3D corrections. We now turn to the results shown in Figure 13 , obtained under assumptions similar to those in Figure 12 , but this time using the He i Stark profiles computed from our computer simulations. We remind the reader that line dissolution is not included in these calculations. This omission largely explains why the spectroscopic masses appear lower at higher temperatures. As already demonstrated in Figure 8 , neglecting line dissolution decreases the \log\textsl{{g}} values by roughly 0.05 dex (or about 0.03 M_{\odot} in mass) for hot ( T_{\rm eff}>15,000 K) DB white dwarfs. Figure 13: Same as Figure 12 but using the He i Stark profiles from the computer simulations of Paper I. The results displayed in Figure 13 obtained from computer simulations are practically equivalent to those shown in Figure 12 , based on the semi-analytical B25 calculations. We are thus forced to conclude that our significant improvements in the microphysics of the helium atom have not resolved the long-standing problem with the spectroscopic mass distribution of DB white dwarfs originally identified by 30 . Furthermore, the inclusion of 3D hydrodynamical corrections fails to reduce the spectroscopic masses in the 24,000~{\rm K}>T_{\rm eff}>17,000 K range, and the discrepancy remains. V Concluding Remarks We presented in this paper improved Stark-broadening calculations for neutral helium lines based on state-of-the-art computer simulations that incorporate corrections for ion dynamics, transitions of electron contributions from the core to the wings of the profile, numerical integration of the time-evolution operator for helium perturbed by a fluctuating electric field, the Debye correction for charge–motion correlations, local density variations, and particle reinjection. From a theoretical standpoint, these line profiles represent a major advance over the semi-analytical B97 profiles used in all previous studies. In those earlier calculations, electron broadening was treated using the impact approximation near the line core and the one-electron approximation in the far wings, with an interpolation scheme applied in an arbitrary fashion to ensure a smooth transition between the two regimes. In our simulations, this transition arises naturally from the underlying physics. A key improvement in our approach is the explicit inclusion of ion dynamics, an effect completely neglected in the semi-analytical method. While ion dynamics has no detectable impact in the context of white dwarfs, it plays an important role in lower-gravity stars, as demonstrated in Figures 20 and 21 of Paper I, where the limitations of the quasi-static approximation are clearly illustrated. In 56 , we introduced new computer-simulation calculations of Stark-broadened He i profiles that unified the treatment of ions and electrons, but severe numerical issues prevented us from presenting extensive tables. These problems have now been overcome thanks to the new numerical procedure described in detail in Paper I. The central improvement is the use of the power-spectrum formalism for line-profile generation, rather than the autocorrelation function used previously. As a result, detailed calculations are now available for the most important optical He i lines, over a wide range of temperatures and electron densities. A major motivation behind this project was to address the discrepancy illustrated in Figure 3 : the photometric mass distribution remains centered near the canonical value of 0.6 M_{\odot} , whereas the spectroscopic mass distribution deviates significantly from this value over several temperature ranges—even after excluding the coolest objects (shown in red) whose very weak helium lines (He i \lambda 4471 equivalent width smaller than 5 Å) render spectroscopic determinations unreliable. These comparisons point to a fundamental issue with the spectroscopic technique, for which detailed He i Stark profiles are a crucial ingredient. Our final results, summarized in Figures 12 and 13 , indicate that the discrepancy persists, most notably in the range 24,000~{\rm K}>T_{\rm eff}>17,000~{\rm K} , regardless of the Stark-broadening theory adopted, the treatment of van der Waals broadening at low temperatures, or the inclusion of 3D hydrodynamical effects. We also demonstrated that the use of SDSS spectra—known to be problematic in earlier work—is not responsible for the discrepancy. A different explanation must therefore be sought. Given that the results shown in Figures 12 and 13 are quite comparable (see also Figure 9 ), it is likely preferable, for the purpose of atmospheric analyses of white dwarfs, to continue relying on our semi-analytical Stark profiles, which incorporate line dissolution—a non-negligible effect at high temperatures, as illustrated in Figure 8 . The impact on spectroscopic masses can be inferred by comparing Figures 12 and 13 . Although the results of our computer simulations may appear discouraging in this context, it is important to recognize that they have enabled substantial improvements to our semi-analytical profiles. Chief among these is our revised treatment of Doppler broadening, which has significant consequences at the cool end of the DB sequence (see Figure 5 ), where the treatment of broadening by neutral particles—still a poorly understood mechanism—also plays an important role (see Figure 12 ). Our simulations additionally revealed several issues in the semi-analytical calculations, including improper frequency sampling and normalization. Finally, they may eventually lead to a more physically motivated interpolation scheme between the one-electron and impact regimes, as briefly discussed in Paper I. Despite the progress presented here, several improvements to the Stark-broadened He i profiles are still possible. For instance, while staying in the semi-classical approximation: (1) highest order terms in the multipole expansion of the potential, (2) mixing of upper states with different principal quantum number n , and of course, (3) quantum effects such as penetrating collisions ( 36 ; 18 ) . However, such refinements are probably second-order and are unlikely to resolve the persistent spectroscopic mass discrepancy for DB white dwarfs in the 17,000–24,000 K temperature range. Likewise, significant further improvements from 3D hydrodynamical models are improbable. Other explanations must be considered, possibly involving the temperature and pressure structures themselves—such as missing opacity sources, or known sources that remain poorly modeled, including resonance lines in the UV (see Chapter 6 of 6 ) or the pseudo-continuum opacity associated with dissolved levels of helium, a process usually neglected for the ground state (in hydrogen as well). On a more encouraging note, our corrected treatment of Doppler broadening substantially mitigates the high- \log g problem commonly reported at the cool end of the DB sequence (see Figures 12 and 13 ), although progress in modeling the broadening by neutral particles remains urgently needed. We thank A. Bédard for a careful reading of our manuscript and his constructive comments, and O. Vincent for helping us with the extraction of the relevant SDSS photometric and spectroscopic data used in this analysis. This work was supported in part by the NSERC Canada and by the Fund FRQNT (Québec). Data availability Both the semi-analytical (with and without line dissolution) and simulation He i line profiles are available on Zenodo, which can be accessed via the following link https://doi.org/10.5281/zenodo.18722143 (catalog doi:10.5281/zenodo.18722143) .