SPT-3G 2025 Results Confirm the Hubble Tension: Deepest CMB Power Spectra

Introduction to the Hubble Crisis

Instrumentation and Observations

1. The SPT-3G Receiver and Focal Plane

   The technological leap achieved by the SPT-3G instrument is fundamentally rooted in its massively scaled focal plane. The receiver houses ~16,000 transition-edge sensor (TES) bolometers, an order of magnitude increase over its predecessor, SPTpol. These sensors are maintained at a cryogenic operating temperature of roughly 250 millikelvins, exploiting the steep resistance curve of superconducting materials poised exactly at their critical transition temperature [cite:102]. This extreme thermal sensitivity allows the detectors to register miniscule fluctuations in the incident microwave radiation.

   To effectively separate the primordial cosmological signal from local astrophysical foregrounds—such as thermal dust emission and synchrotron radiation from our own Milky Way—the focal plane is populated with trichroic pixels. Each pixel simultaneously couples to radiation at 95, 150, and 220 GHz. This multifrequency capability is critical for cleaning the maps, as the spectral index of galactic dust differs significantly from the blackbody spectrum of the CMB, allowing for highly efficient component separation algorithms to isolate the true cosmological temperature and polarization anisotropies.

2. The 1,500 deg² D1 Survey Field

   The observational campaign detailed in Camphuis et al. 2025 focuses on the D1 survey field, a meticulously chosen 1,500 square degree region of the southern sky. This specific footprint was selected for its exceptionally low foreground contamination and its geometric overlap with other premier cosmological surveys, including the Dark Energy Survey (DES) and the BICEP/Keck array. Observing continuously from the geographic South Pole, the telescope benefits from an exceptionally stable, dry atmosphere, minimizing absorption and emission from atmospheric water vapor.

   The integration time dedicated to the D1 field has resulted in the deepest ever TT, TE, and EE maps over this angular scale. The noise levels in the 150 GHz band, where the CMB signal-to-noise is maximized, have reached unprecedented depths of just a few microkelvin-arcminutes. This raw sensitivity is particularly crucial for measuring the E-mode polarization (EE) and the temperature-E-mode cross-correlation (TE), both of which contain sharper acoustic peak structures and are far less contaminated by extragalactic foregrounds than the traditional temperature (TT) maps.

Data Processing and Power Spectra

1. Mapmaking and Systematics Mitigation

   Transforming the raw, time-ordered data (TOD) from 16,000 detectors into pristine sky maps requires a sophisticated, supercomputing-intensive pipeline. The initial stage involves aggressive temporal filtering to remove low-frequency 1/f noise induced by atmospheric turbulence and instrumental thermal drifts. A polynomial subtraction is applied across the detector arrays, effectively acting as a high-pass filter. While this removes large-scale noise, it also inevitably filters out some of the true large-scale cosmological signal, an effect that must be carefully quantified and corrected for via the transfer function during the power spectrum estimation phase [cite:103].

   Following calibration against celestial standards and internal thermal sources, the TOD is projected onto a HEALPix grid to produce Stokes I, Q, and U maps. The Q and U maps, which describe the linear polarization state of the incoming light, are subsequently decomposed into curl-free E-modes and divergence-free B-modes. The SPT-3G pipeline employs rigorous null tests—differencing maps made from data split by time, telescope scan direction, and focal plane properties—to ensure that residual systematics, such as beam asymmetries or detector crosstalk, are well below the statistical noise floor of the survey.

2. Angular Power Spectrum Extraction

   The cosmological information contained in the sky maps is compressed into the angular power spectrum, which quantifies the variance of the temperature and polarization fluctuations as a function of angular scale. Due to the incomplete sky coverage of the D1 field, the true power spectrum cannot be computed via a direct spherical harmonic transform. Instead, the collaboration utilizes a pseudo-C_ℓ estimator, applying an apodized mask to smoothly down-weight the noisy map edges and point sources.

   D_ℓ = ℓ(ℓ+1) ⟨|a_ℓm|²⟩ / 2π

   This formulation corrects for the mode-coupling introduced by the survey mask, allowing the recovery of the unbiased bandpowers. The resulting TE and EE spectra from the 2025 data release resolve the acoustic peaks up to a multipole moment of ℓ ≈ 3000 with astonishing precision. The damping tail—the exponential suppression of power at small angular scales due to photon diffusion during recombination—is mapped better than ever before, providing immense leverage on fundamental parameters such as the primordial helium abundance and the physical baryon density.

Cosmological Parameter Estimation

1. Baseline ΛCDM Fits and Structure Growth

   To extract the cosmological parameters, the SPT-3G bandpowers are fed into a Markov Chain Monte Carlo (MCMC) likelihood analysis, exploring the high-dimensional parameter space of the baseline ΛCDM model. The model relies on six free parameters: the physical baryon density, cold dark matter density, angular size of the sound horizon, optical depth to reionization, and the amplitude and spectral index of the primordial scalar perturbations. The SPT-3G data alone provide incredibly tight constraints on these parameters, entirely consistent with the broader framework of a dark energy-dominated, flat universe.

   A particularly intriguing result from the 2025 release is the measurement of the amplitude of matter clustering in the late universe, parameterized by σ8. The SPT-3G data yield σ8 = 0.8137, a value derived from the early-universe amplitude extrapolated forward in time. However, recent large-scale structure measurements from the DESI DR2 survey, which directly probe the late universe, prefer a lower value for the clustering amplitude. The comparison between the SPT-3G early-universe prediction and the DESI DR2 late-universe measurement reveals a 2.8σ tension, hinting at a potential suppression of structure growth that ΛCDM fails to predict [cite:104].

2. The Hubble Constant Constraint

   The crown jewel of the Camphuis et al. 2025 study is its constraint on the Hubble constant. By accurately measuring the angular scale of the acoustic peaks and the physical size of the sound horizon at recombination, the CMB provides a geometric standard ruler. Calibrating this ruler against the tightly constrained matter densities allows for a robust determination of the expansion rate required to project that physical scale onto the observed angular scale. Using only the SPT-3G D1 data, the collaboration derives H0 = 66.66 ± 0.60 km/s/Mpc.

   To maximize statistical power, the researchers combined the SPT-3G data with the overarching CMB Spectral Parameter Analysis (CMB-SPA) framework, which harmonizes data from Planck, the Atacama Cosmology Telescope (ACT), and SPT. This joint early-universe likelihood yields a highly formidable constraint of H0 = 67.24 ± 0.35 km/s/Mpc. The sub-percent precision of this measurement effectively closes the door on the hypothesis that the low H0 values derived from the CMB are merely the result of a single experiment's hidden systematic errors.

Implications for the Hubble Tension

Conclusion