Probing the Shock Breakout Signal of SN 2024ggi from the Transformation of Early Flash Spectroscopy

Utilizing the Li-Jiang 2.4-m telescope (LJT; Fan et al., 2015) equipped with YFOSC (Yunnan Faint Object Spectrograph and Camera; Wang et al., 2019), we obtained a classification spectrum for SN 2024ggi at $\sim$ 13.9 hours after its explosion (Zhai et al., 2024). This classification spectrum was predominantly characterized by narrow emission lines of H, He, and CNO elements, as seen in Figure 1. These features, also known as flash features (Gal-Yam et al., 2014), are commonly observed in SNe IIn (e.g., SN 1998S, Leonard et al., 2000). They are generated by recombinations of the surrounding dense CSM that is photoionized by radiation from the embedded shock.

Interestingly, in the initial-phase spectrum, SN 2024ggi exhibits a significant difference from other SNe II-P/CSM events like SN 2013fs, SN 2018zd, and SN 2023ixf. As presented in Figure 2 (a), the classification spectrum of SN 2024ggi lacks strong emission features of high ionization states. It showed lines of He i, N iii, and C iii rather than those of He ii, N v and C iv or highly-ionized O v $\lambda$$\lambda$ 5576,5598 and N v $\lambda$$\lambda$ 4604,4620 lines that appeared in SN 2013fs.

In particular, SN 2024ggi and SN 2013fs exhibit two opposite spectral line morphologies in the 4600-4700 Å region. The spectra of SN 2013fs at $t\leq 10$ hr (where $t$ denotes time after the first light) show a narrow He ii $\lambda$ 4686 emission line, while the nearby N iii $\lambda$$\lambda$ 4630,4641 doublets appear flatter without prominent narrow-line components. Instead, there is a small bump consisting of N v $\lambda$$\lambda$ 4604,4620. However, the classification spectrum of SN 2024ggi reveals narrow emission lines of N iii $\lambda$$\lambda$ 4630,4641, while the position corresponding to He ii $\lambda$ 4686 appears flat. These two SNe belong to those with the earliest discoveries among the sample of SNe II-P with CSM. Spectroscopic data from other samples, usually obtained one day after the SBO, did not show similar phenomena, suggesting a high dependency of early-time spectra on the exact post-breakout epoch and the CSM properties.

Given the observed IIn-like features in the classification spectrum and a relatively lower ionization state revealed by these features, we initiated an hour-cadence observation campaign with the LJT. A total of four spectra were acquired within 2.4 hours after the identification, which allowed us to monitor possible variations in the ionization state for this young SN II.

From the second spectrum, SN 2024ggi exhibited narrow He ii $\lambda$ 4686 emission that was absent before. It is possible that, in the first spectrum, the CSM has not been shocked enough to generate sufficient heat for emitting a significant amount of He ii, except for the deep CSM near the shock, where the material could be rushing. The subsequent two spectra did not show significant morphological evolution. Further analysis of the equivalent widths (EWs) of the main spectral lines, as shown in Figure 3, reveals that within the 1.5 hr from $t\sim$14.7 hr to $t\sim$ 16.2 hr, the strength of He ii $\lambda$ 4686 and He i $\lambda$ 6678 lines remained almost unchanged. In contrast, the C iii $\lambda$ 5696 and He i $\lambda$ 7065 lines beaome gradually weak.

On the second day after explosion, an observation relay across different time zones was conducted with LJT, Telescopio Nazionale Galileo (TNG), and Very Large Telescope Unit 1 (VLT UT-1), which allowed a continuous monitoring of the rapid evolution of SN 2024ggi. In particular, to examine the structure of narrow spectral lines, we utilized the cross-dispersion capability of YFOSC with a resolution of 3500. This resolution had proven its effectiveness in our previous research of SN 2023ixf, enabling us to discern the intricate structure of H$\alpha$ and provide a more precise constraint on its broadening (Zhang et al., 2023).

Nonetheless, in the case of SN 2024ggi, the H$\alpha$ line had already undergone significant broadening in the spectra taken at $t\sim 38.7$ hr, with a full width at half maximum (FHWM) of $\sim$ 700 km s^-1, exceeding the instrumental FWHM (i.e., $\sim$ 85 km s^-1). Concurrently, as the narrow line broadened swiftly, its equivalent width (EW) decline steeply at $t\gtrsim 30$ hr, as illustrated in Figure 3.

The rapid broadening and decreasing of the H$\alpha$ narrow component testifies the rapid spectral evolution of SN 2024ggi. By $t\sim$ 29 hr, many highly ionized spectral lines emerged in the spectrum, including O iv, N iv, and C iv, and even O v. Due to limitations in the signal-to-noise ratio (S/N) of the spectrum, the He ii $\lambda$ 5411 line can only be marginally detected at $t<1$ d, but it becomes visible at $t\sim$ 29 hr. The most notable change is that One day after explosion, the emission lines of He i $\lambda$ 7065 and C iii $\lambda$ 5696 disappeared in the spectra, but replaced by C iv $\lambda$$\lambda$ 5801, 5811, C iv $\lambda$ 7110 and N iv $\lambda$ 7122 lines. As shown in Figure 3, He ii $\lambda$ 4686 continues to weaken throughout the day, while the C iv $\lambda$$\lambda$ 5801, 5811 progressively gain the strengthens. We notice that the narrow He i $\lambda$ 6678 line disappeared at 29.2 hr but it reappeared five hours later. The He i $\lambda$ 5876 line is not detected perhaps due to that it coincides with the Na i D absorption of the Milky Way.

At $t\sim 29$ hr, the spectral line flux and profile morphology in this region are unreliable due to a saturation in the wavelength range from $\sim$4610Å to $\sim$4700 Å. Therefore, we cannot confirm whether the appearance of N v at this time is genuine, nor can we ascertain the reliability of the intensity contrast between N iii and He ii. At $t\sim$ 34 hr, the narrow lines of He ii $\lambda$ 4686 and N iii $\lambda$$\lambda$ 4630,4641 have comparable intensities and thus formed a double peak profile. Such a double-peak structure has been also seen in some SNe II. For instance, it appeared in SN 2023ixf at $t\sim$ 1.2 d, SN 1998S at $t\sim$ 2.3 d, and SN 2018zd at $t\sim$ 4 d. SN 2013fs may also exhibit a similar structure at $t\sim$ 1.4 d, except that the left peak corresponds to N v $\lambda$$\lambda$ 4604,4620 instead of N iii in its spectrum (Dessart et al., 2017; Yaron et al., 2017). The N v doublet is also visible in the spectra of SN 2024ggi taken at $t=$ 29.2h, 38.3h, and 38.7h. This indicates that SN 2013fs and SN 2024ggi can reach similar ionization levels, though they may have different shock emissions and CSM environments.

At $t>$ 38 hr, N iii doublet becomes weaker than He ii $\lambda$ 4686 in SN 2024ggi. However, the N iii doublet remains unchanged over the following days while He ii  shows a faster decline in intensity. Overall, all narrow emission lines become less distinct after the third day, as they broaden significantly with a velocity exceeding 1000 km s^-1. The disappearance of narrow emission lines indicates that the IIn-like phase is ending and SN 2024ggi is gradually entering a new stage of evolution.

The rapid spectral evolution relates in part to a temperature change. We can roughly estimate the temperature of the SN photosphere through fits to the spectral energy distribution (SED) assuming a blackbody emission. In Figure A1, the temperature of SN 2024ggi is inferred to be about 13500 K during the phase from 13.9 hr to 16.2 h, and it increases sharply from $\sim$ 13500 K to $\sim$ 26000 K about half a day later (also reported by Chen et al., 2024b). The rapid temperature rise is concomitant with the rapid increase in ionization, and both are attributed to the photoionizing radiation from the shock.

Since SN 2024ggi exhibited a rapid evolution, we maintained frequent monitoring until the third day, gradually tapering off the pace after that. As presented in Figure 1, we can only see a few weak ‘narrow’ emission lines, such as H$\alpha$ and N iii $\lambda$$\lambda$ 4634,4641 doublet at $\sim$ 3.6 d. Meanwhile, P-cygni profiles of H$\beta$, H$\gamma$, and He i $\lambda$ 5876 begin to emerge.

At $3<t<15$ d, the spectral lines of SN 2024ggi undergo a significant transition. Initially, these lines exhibited broadening due to electron scattering, formed within the unshocked CSM. However, they gradually shift to Doppler broadening as they initially formed within the fast-moving dense shell and subsequently arises also from the ejecta as the dense shell gradually becomes optically thinner. This evolution is accompanied by a noticeable blueshift in their emission peaks. Additionally, these lines begin to appear as P-Cygni profiles, with both absorption and emission components increasing in strength over time.

Roughly two weeks after the explosion, SN 2024ggi developed spectral features typical of SNe IIP in both optical and NIR spectra, i.e., the appearance of broad H i and He i lines. For example, as seen in FigureA2, the NIR spectrum of SN 2024ggi, obtained with the 6.5 m Magellan telescope equipped with FIRE (Folded-port InfraRed Echellette) at t$\sim$ 14 d, reveals similar spectral features as seen in SN2017eaw (Szalai et al., 2019).

As observed in SN 2018zd (Zhang et al., 2020) and SN 2023ixf (Zhang et al., 2023), the SN-CSM interaction creates a slow evolution in spectral features after the narrow, electron-scattering broadened emission lines disappear (Figure 2). During shock wave propagation, the CSM is shocked, and compressed into a dense shell, converting a fraction of kinetic energy into thermal energy and heating the CSM as well as boosting the SN luminosity. In the case of SN 2024ggi, photoionization predominantly facilitates this heating process within the first three days. Consequently, emission from the post-shock gas, such as the cold dense shell (CDS), originates from the release of shock-deposited energy (i.e., deposited at earlier times). The relatively faint spectral lines observed approximately five days later are likely attributed to the spectrum primarily forming within the CDS during this period. This weak emission is indicative of a steep density gradient of the CDS (Dessart et al., 2017).

We note that the evolution speed of the photospheric features in SN 2024ggi is slower than that of the typical SN 1999em but still faster than in SN 2018zd and SN 2023ixf. As illustrated in Figure 2 (b), SN 2024ggi evolves into spectra with prominent P-Cygni profiles at around $t\sim$ 23 d, and its spectrum at $t\sim 32$ d shows a close resemblance to that of a regular SN IIP. Nevertheless, SN 2024ggi evolves faster in the first month, regardless of early IIn or later photospheric spectra, compared to SN 2018zd and SN 2023ixf, suggesting a more compact CSM.

Within the initial 72 hr after the first light, we obtained a total of twelve spectra revealing rapid evolution of line strengths, in particular for lines associated with different ionization levels (e.g., He i vs He ii). During this period, the lower ionization species (such as He i, C iii, and N iii) gradually weakened and disappeared, while the strength of emission lines from higher-ionization species (like He ii, N iv, and C iv) increased (see also Section 2.2).

The evolution of these line fluxes or EWs is however more complex. The spectral lines such as He i $\lambda$ 6678, He i $\lambda$  7065, and C iii $\lambda$ 5696 disappeared at $t\sim$ 29.2 hr, while higher ionization lines, including C iv $\lambda$$\lambda$  5801,5811 and O v $\lambda$$\lambda$ 5576,5598, emerged. However, five hours later, He i $\lambda$ 6678 reappeared, while O v became more elusive. In the 38.3 hr and 38.4 hr spectra, roughly four hours afterward, He i $\lambda$ 6678 vanished once again, O v reappeared, while N iii significantly diminished, and N v became detectable.

The weakening of N iii and the emergence of N v were already evident in the 29.2 hr spectrum. Considering the evolution of O v  and He i, we believe the relative intensities of N v and N iii observed at 29.2 hr are reliable despite the detector saturation in this wavelength region. Thus, we observed a decreasing and increasing ionization across the 29.2, 34.4, and 38.3 hr spectra.

This trend is mirrored in Figure 3, where a similar fluctuation in C iv $\lambda$$\lambda$ 5801,5811 is apparent in $t\sim$ 30 h spectrum. These variations suggest that although the ionization evolution initially arose and fell within the first 72 hr, there were fluctuations when the ionization reached its maximum at around 30 hours after the first light. These fluctuations are probably related to the complicated propagation through a clumpy, and perhaps asymmetric CSM. The alteration in the ionization state indicates the complex interaction between the shockwave, the surrounding material and the high-energy radiation environment in the early stages of SN explosion.

To further understand the evolution of the ionization state of SN 2024ggi, we compare the observed spectra with the spectral models presented in Dessart et al. (2017) (referred to as D17 below), as shown in Figure 4. This model set successfully reproduces the observational manifestations of SNe II, specifically those featuring short-lived IIn-like spectra while taking into account the varying physical conditions of the progenitor before the explosion. The methodology adopted in D17 involves creating a comprehensive grid of models for different CSM characteristics (including radius, mass, density, density profile, velocity, and composition) based on the same progenitor and explosion parameters, thereby encompassing a wide range of the parameter space.

Although none of the models in D17 can fully reproduce the entire spectral features, models like r1w5r and r1w6 provide a good reference for us to track and study the ionization state and evolution of the CSM. In comparison, we found that the r1w5r model fits the observations well, especially at $t\sim 1.5$ d, when it can reproduce all the observed characteristics of SN 2024ggi. Moreover, at $t\sim 1.17$ d, high-ionization lines such as O iv and N iv appeared in the model spectrum, which is comparable to the spectrum of SN 2024ggi at $t\sim$ 1.35 d, indicating that the ionization states of SN 2024ggi is consistent with the model during this period. However, the CSM in the r1w5r model has a higher ionization state at $t<1$ d than that observed in SN 2024ggi. For example, the r1w5r model does not show any He i line in the early stage, and in fact No He i line appears in any of the D17 models¹¹1The reason is partly because no model was computed during the earliest epochs when the first radiation from the shock was crossing the CSM (i.e., the radiative precursor). Modeling this phase is quite challenging, and there are inconsistencies in the physics addressed in D17. One such inconsistency relates to light travel time effect, considered in radiation-hydrodynamics calculations but omitted during the post-processing phase of radiative transfer calculations for spectra. . In contrast, the He ii line in the r1w5r model is much stronger than the N iii line at the beginning of explosion.

Interestingly, the blue edge of H$\alpha$ is quite similar in the $t\sim 16$ spectrum of SN 2024ggi and the $t=$ 15 d r1w5r model spectrum. The only difference lies in the emission strength and the extent of the profile to the red. One reason might be that in the model, the optical depth attenuation is much more significant, which suggests the CDS breaks up and becomes clumpy significantly. This could be resolved if the CDS was allowed to break-up, as would occur in multi-D radiation-hydrodynamics with the development of Rayleigh-Taylor instabilities.

Jacobson-Galán et al. (2024b) favored the r1w6 model over the r1w5r model in their comparisons. The mass loss rates of these two models are quite similar, and their early spectra are also very similar. Both models exhibit a higher ionization in the early stages than observed in SN 2024ggi. The early He ii line in the r1w6 model is stronger than that in the r1w5r model, leading us to believe that the r1w5r model performs better in this aspect. Additionally, since the spectra of Jacobson-Galán et al. (2024b) do not extend beyond ten days after the explosion, they did not observe the distinct P-cygni features that appeared later in SN 2024ggi. It is worth noting that the early stage spectra of the r1w6 model (up to 14 days) also do not show this characteristic, making it inferior to the r1w5r model.

The r1w5r model is suitable for SN 2024ggi, as it can roughly reproduce the observed main IIn-like spectral features. However, the issue lies in the fact that the early ionization state of the model is too high. The CSM property of SN 2024ggi might be intermediate between r1w5r and r1w6. There may also be additional features ignored in D17, such as asymmetry, the break-up of the CDS, clumping in the CSM, and such. And the radiative transfer modeling could be improved. Therefore, the initial spectral evolution of SN 2024ggi provides a new opportunity for future developments of spectroscopic models.

In the early phase, when the SN photosphere is within the unshocked ionized CSM, the spectra are characterized by symmetric emission lines with narrow cores and extended, electron-scattering broadened wings. Figure A3 shows the double-component fitting of the early H$\alpha$ emission of SN 2024ggi. Note that the broader wings are referred as mid-width components to distinguish them from the even wider emission lines in the later P-Cygni profiles Doppler-broadened in the fast-expanding ejecta.

Based on the two-component fitting of Figure A3, we obtain the parameters and evolution of each component. To visually compare this evolution in H$\alpha$, we selected spectra taken by YFOSC+G14 in the first three days, as presented in the top panel of Figure 5, to minimize the influence of instrumental effects. All of these spectra are corrected for the redshift and rotation of the host galaxy; see more detail in section A.2.

Given the low spectral resolution, an instrumental correction is necessary to accurately determine the FWHM of the H$\alpha$ line via FWHM_cor = (FWHM${}^{2}_{\rm obs}$ $-$ FWHM${}^{2}_{\rm inst})^{1/2}$, where instrumental FWHM is estimated through the skyline O i $\lambda$ 6300.3. This method has been approved effective in the study of SN 2023ixf, where the value we obtained through mid-resolution spectra (Zhang et al., 2023) matches well with that from the high-resolution spectra (Smith et al., 2023), as seen in the middle panel of Figure 5. In this panel, the corrected FWHM of the narrow component broadens from about 50 km s^-1 to approximately 200 km s^-1 within hours on the first day. Initially, the FWHM_obs is close to the instrumental FWHMM_inst, making corrections potentially inaccurate. We propose 50 km s^-1as an upper limit of FWHM before radiation broadening, and the actual FWHM is likely narrower at $t<14$ hr.

Despite a measurement uncertainty of around 50% in the third spectrum, there is a discernible broadening of spectral lines, indicating the onset of electron scattering effects as well as the potential radiative acceleration of the unshocked CSM (D17). At this point, using the FWHM of the narrow lines to constrain the original CSM velocity becomes unfeasible.

Pessi et al. (2024) measured an FWHM $\approx$ 55 km s^-1 of H$\alpha$ at $t\sim$ 23.6 hr (the epoch is calculated by the first light adopted in this letter) with high-resolution spectroscopy. They found that this line broadened to 61 km s^-1 seven hours later. Their measurements indicate that although the low-resolution results are not precise enough, they can provide certain constraints at the earliest hours.

At $t\sim$ 38.7 hr, a mid-resolution spectrum revealed that the FWHM of H$\alpha$ had increased significantly, rendering instrumental broadening negligible. Remarkably, the narrow spectral lines of SN 2024ggi expanded to over 700 km s^-1 within 48 hr. In contrast, the H$\alpha$ FWHM of SN 2023ixf during the similar timescale after the first light was only 55 km s^-1 ($t\sim 43$ hr, Zhang et al., 2023), see in the mid-panel of Figure 5. Furthermore, the broadening of the H$\alpha$ narrow-line component in SN 2023ixf occurred much slower, increasing only from 55 km s^-1 to 140 km s^-1 within the first three days, and reaching close to 900 km s^-1 when the narrow line was almost disappearing on the fifth day. Similarly to SN 2023ixf, the FWHM of the narrow H$\alpha$ of SN 2020pni increased from 250 km s^-1 to nearly 1000 km s^-1 in the first five days (Terreran et al., 2022). The more rapid and drastic broadening observed in SN 2024ggi suggests a less extended CSM than that of SN 2023ixf and SN 2020pni.

Although we cannot use the FWHM to limit the CSM velocity of SN 2024ggi, a noticeable blueshift was observed in the H$\alpha$ emission. This shift is evident in the top panel of Figure 5, with precise measurements detailed in the middle panel. After some velocity adjustments and a refined wavelength calibration using the skyline O i$\lambda$ 6300.3, an initial blueshift velocity of $-12\pm 20$ km s^-1 was measured at $t\sim 13.9$ hr. The resolution of the second spectrum at $t\sim 14.7$ hr is slightly higher, and the measured velocity is $-20\pm 10$ km s^-1. Considering the measurement errors in galactic rotation, the lower limit of the stellar wind velocity observed at the outer layer of CSM around SN 2024ggi is $20\pm 15$ km s^-1. A fast expansion of narrow H$\alpha$ component was also seen in the SNe II with a short-lived IIn-like feature, e.g., SN 2020pni

We averaged the measurements daily to reduce uncertainties, yielding mean velocities of $42\pm 29$ km s^-1, $99\pm 23$ km s^-1, and $133\pm 17$ km s^-1  in the first three days, respectively. These measures reveal the acceleration of the unshocked CSM by the radiation from the shock. The acceleration implies that the earlier observations can more genuinely reflect the movement of CSM before the explosion, representing the progenitor’s stellar wind speed in its final phase. Thus, the inferred CSM velocity of SN 2024ggi does not exceed 40 km s^-1 consistent with the high-dispersion observation results (e.g., 37 km s^-1, Shrestha et al., 2024). Given the lower limit estimated before, the stellar wind velocity derived by our initial observation suggests that the progenitor wind velocity of SN 2024ggi is between 20 and 40 km s^-1.