Decay of superheavy nuclei based on the random forest… — Plain-Language Explanation

✨

This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the periodic table of elements as a vast, crowded city. Most of the buildings (elements) we know are stable and safe. But as you move to the very edge of the city, into the "Superheavy District," the buildings become incredibly unstable. They are like houses made of Jenga blocks that are about to collapse at any moment.

Scientists want to build new buildings in this district (elements 119, 120, 121, etc.) and find a "Stability Island"—a place where these wobbly houses might actually stand still for a while. But to find them, they need to know: How will these new buildings fall apart? Will they crumble slowly (Alpha decay)? Will they explode instantly (Spontaneous Fission)? Or will they transform into a different building entirely (Beta decay)?

This paper is about using a digital crystal ball (Machine Learning) to predict exactly how these super-heavy atoms will decay.

The Problem: Old Maps vs. New Territory

Traditionally, physicists used "semi-empirical formulas." Think of these as old, hand-drawn maps. They work great for the parts of the city we've already explored, but when you try to use them in the uncharted Superheavy District, the maps start to get fuzzy, contradictory, or even point in the wrong direction. They are like trying to navigate a new city using a map from 50 years ago; the roads have changed, and the old signs don't match reality.

The Solution: The "Random Forest" Algorithm

Instead of relying on one old map, the authors used a Random Forest algorithm.

The Analogy:
Imagine you are trying to guess the weather in a new city.

The Old Way: You ask one expert meteorologist who uses a single complex formula. If their formula is slightly off, your prediction is wrong.
The Random Forest Way: You ask 100,000 different meteorologists (trees).
- Each one looks at a slightly different set of clues (some look at humidity, some at wind, some at pressure).
- Each one makes their own prediction based on a random sample of past weather data.
- Then, you take the average of all 100,000 predictions.

By averaging out the mistakes of the individual "trees," the forest becomes incredibly accurate and robust. It doesn't get confused by outliers or weird data points. In this paper, the "clues" are the number of protons and neutrons in the nucleus, and the "prediction" is how long the atom will last before decaying.

What Did They Find?

1. The "Alpha Decay" Highway
For most of the new, heavy elements they looked at (specifically elements 119, 121, and their neighbors), the forest predicts that Alpha Decay will be the main way they fall apart.

Analogy: It's like a slow leak in a tire. The nucleus sheds a small piece (an alpha particle) and becomes a slightly lighter, more stable element. This is the "dominant signal" scientists should look for when trying to identify these new elements.

2. The "Spontaneous Fission" Explosion
However, there is a catch. For some specific even-numbered elements (like 120 and 122), the nucleus might not just leak; it might explode (Spontaneous Fission).

The Odd-Even Effect: The paper highlights a weird quirk of nature. Nuclei with even numbers of protons and neutrons are like a perfectly balanced stack of blocks—they are more likely to collapse suddenly (fission) than their neighbors with odd numbers. This "odd-even staggering" means that for elements 120 and 122, scientists must be ready for a sudden explosion, not just a slow leak.

3. The "Long-Lived Island" in the Southwest
The most exciting discovery is a predicted "island of stability" located at a specific spot (around element 114, neutron number 184).

Analogy: Imagine a stormy sea where almost every boat sinks quickly. But in one specific corner (the Southwest), there is a calm lagoon where boats can float for days or years.
The paper predicts a "long-lived circle" here. Because of a balance between the forces holding the nucleus together and the forces pushing it apart, these specific atoms might last much longer than their neighbors. This is the "Holy Grail" for nuclear physicists.

4. The "Black Box" Candidates
The algorithm also pointed out several specific atoms (like certain isotopes of Curium, Einsteinium, and Mendelevium) that haven't been measured yet but are predicted to be surprisingly long-lived.

Analogy: It's like a treasure map marking spots where gold might be buried, even though no one has dug there yet. The authors suggest these are the best candidates for future experiments.

Why Does This Matter?

Finding new elements is like trying to catch a ghost. They appear for a split second and vanish. To catch them, you need to know exactly how they vanish so you can set up your detectors to catch the "ghost's trail."

This paper tells us:

Where to look: Focus on the Southwest corner of the chart.
What to look for: Mostly alpha decay (the slow leak), but watch out for spontaneous fission (the explosion) in even-numbered elements.
How to look: Use the Random Forest method to filter out the noise of old, inaccurate formulas.

In short, the authors built a super-smart digital guide to help us navigate the chaotic edge of the atomic world, pointing us toward the new elements that might finally reveal the secrets of the "Island of Stability."

1. Problem Statement

The identification and characterization of new superheavy elements (SHEs) beyond Oganesson ( $Z=118$ ) and the search for the "island of stability" are hindered by the extreme instability of these nuclei, which often possess very short half-lives. A critical challenge in nuclear physics is predicting the dominant decay mode (e.g., $\alpha$ -decay, $\beta$ -decay, spontaneous fission) and the corresponding partial half-lives for unknown nuclides.

Traditional approaches rely on:

Microscopic theories: Based on nucleon-nucleon interactions, often computationally expensive.
Macroscopic semi-empirical formulas: Fit to experimental data but suffer from overfitting, inappropriate parameterization, and significant divergence when extrapolating to unmeasured regions (especially for Spontaneous Fission, SF).

The authors aim to address the limitations of existing formulas, particularly their failure to accurately describe the competition between decay modes and their divergence during extrapolation, by applying machine learning to correct the residuals of these formulas.

2. Methodology

The study focuses on the nuclear region $Z \ge 84$ and $N \ge 128$ . The methodology involves a hybrid approach combining semi-empirical formulas with the Random Forest (RF) machine learning algorithm.

A. Baseline Semi-Empirical Formulas

The authors first calculate partial half-lives ( $T_{1/2}$ ) for five decay channels using established formulas:

$\alpha$ -decay: Universal Decay Law (UDL).
Spontaneous Fission (SF): A new three-parameter formula (SF3) is proposed to avoid the divergence seen in higher-order polynomial formulas (e.g., Ren, Xu, Santhosh, Soylu) during extrapolation.
$\beta$ -decay ( $\beta^-$ , $\beta^+$ ) and Electron Capture (EC): Formulas based on Gamow-Teller transitions and phase-space factors.

B. Random Forest (RF) Training

Instead of replacing the physical formulas, the RF algorithm is trained to predict the residuals (the difference between the logarithmic experimental half-lives and the calculated values from the formulas).

Input Features: Proton number ( $Z$ ), Neutron number ( $N$ ), Mass number ( $A$ ), parity of $Z$ and $N$ , and decay energy (or Fission Barrier $F_B$ for SF).
Algorithm: An ensemble of 105 Decision Trees (DT) using bootstrap sampling. This approach minimizes overfitting and handles correlations between features effectively.
Correction: The final predicted half-life is the sum of the semi-empirical formula result and the RF-predicted residual.

C. Data Sources

Experimental Data: 445 nuclides with measured partial half-lives and branching ratios from NUBASE2020.
Theoretical Energies: For nuclides without experimental mass data, energies are derived from the WS4 mass model and the UNEDF0 energy density functional.

3. Key Contributions

Development of a Robust SF Formula: The authors propose the SF3 formula, which avoids the mathematical divergence (orders of magnitude errors) observed in other SF formulas when extrapolating to superheavy regions.
Machine Learning Integration: This is the first application of the Random Forest algorithm specifically to study the competition between different decay modes in superheavy nuclei by correcting residuals of partial half-life formulas.
Systematic Prediction: The study provides a comprehensive map of dominant decay modes and half-lives for the superheavy region, including new elements $Z=119-122$ .

4. Key Results

A. Accuracy and Validation

The model successfully reproduces observed half-lives and dominant decay modes.
96.9% of nuclei beyond $^{212}\text{Po}$ have their dominant decay mode correctly described.
The Root Mean Square Error (RMSE) for the dominant decay mode is reduced to 0.62 (using WS4 energies) and 0.67 (using UNEDF0 energies) after RF training.
The model correctly identifies nuclides where two decay modes compete (e.g., $\alpha$ vs. SF in $^{255}\text{Rf}$ , $^{262}\text{Db}$ , $^{286}\text{Fl}$ ).

B. Predictions for New Elements ( $Z=119-122$ )

Dominant Mode: $\alpha$ -decay is predicted to be the dominant mode for isotopes of $Z=119, 120, 121, 122$ (with $N \le 184$ ).
Odd-Even Staggering: A strong odd-even staggering effect is observed in the spontaneous fission half-lives of even- $Z$ isotopes. Even-even nuclei have significantly shorter SF half-lives compared to their odd- $A$ neighbors due to pairing effects.
Competition: For even-even isotopes of $Z=120$ and $122$, Spontaneous Fission becomes competitive with $\alpha$ -decay, whereas $\alpha$ -decay remains the clear signal for odd- $Z$ isotopes ( $Z=119, 121$ ).

C. The "Long-Lived" Island

A region of enhanced stability (longer half-lives) is predicted at the southwest of $^{298}\text{Fl}$ (around $Z=114, N=184$ ).
This stability is attributed to a high fission barrier ( $F_B$ ) caused by the competition between nuclear deformation and Coulomb repulsion.
The study identifies specific candidates for future measurement with predicted half-lives $> 10^4$ $> 1 0^{4}$ s:
- $^{250,252,254}\text{Cm}$
- $^{260,261}\text{Es}$
- $^{261-264}\text{Md}$
- $^{265}\text{Lr}$

D. Comparison with Other Models

The RF-corrected SF3 formula shows significantly better stability in extrapolation compared to Ren2005, Xu2008, Santhosh2010, and Soylu2019, which exhibit divergent behavior (differences of tens of orders of magnitude) far from known shells.
The model captures the experimental odd-even staggering in SF half-lives, which many other theoretical models fail to predict accurately.

5. Significance

Guidance for Experiments: The predictions provide crucial guidance for experimental facilities (such as CAFE2 and SHANS2 in Lanzhou) on which isotopes to target for synthesizing new elements. Specifically, it suggests focusing on $\alpha$ -decay signals for $Z=119/121$ and considering SF competition for $Z=120/122$ .
Understanding Stability: The study highlights that Spontaneous Fission is the limiting factor for the stability of superheavy nuclei. Accurate modeling of SF, particularly beyond $^{286}\text{Fl}$ , is essential for locating the center of the island of stability.
Methodological Advancement: The work demonstrates that machine learning, when used to correct residuals of physically motivated formulas rather than acting as a "black box," offers a robust strategy for extrapolating nuclear properties to unmeasured regions, reducing the risk of overfitting and divergence.

In conclusion, this paper establishes a reliable framework for predicting the decay properties of superheavy nuclei, refining our understanding of the stability landscape, and identifying specific targets for the next generation of superheavy element discovery.

Decay of superheavy nuclei based on the random forest algorithm