diff --git a/make_mass_attenuation.py b/make_mass_attenuation.py
new file mode 100644
index 0000000..6fe7793
--- /dev/null
+++ b/make_mass_attenuation.py
@@ -0,0 +1,99 @@
+#!/usr/bin/env python
+"""Obtain photon mass attenuation coefficients for elements Z=1 to 100 from the
+NIST Standard Reference Database 8 (XCOM Photon Cross Sections Database,
+https://www.nist.gov/pml/xcom-photon-cross-sections-database) and write them
+to an HDF5 file.
+
+Data are retrieved by submitting a POST request to the XCOM CGI for each
+element. The resulting HDF5 file contains 100 datasets (keyed by zero-padded
+atomic number, e.g. '001'--'100'), each a 2D array with shape (2, N) where
+row 0 is photon energy in [eV] and row 1 is the mass attenuation coefficient
+mu/rho (Total Attenuation with Coherent Scattering) in [cm^2/g].
+"""
+
+import time
+from urllib.parse import urlencode
+from urllib.request import urlopen, Request
+
+from lxml import html
+import numpy as np
+import h5py
+from openmc.data import ATOMIC_SYMBOL
+
+
+XCOM_URL = 'https://physics.nist.gov/cgi-bin/Xcom/xcom3_1'
+
+# POST payload template
+# OutOpt='PIC': all quantities in cm2/g
+# Output='on':  include the standard energy grid
+# NumAdd='1':   number of additional energy input rows (hidden field default)
+# Graph0='on':  select "None" for graph (faster; data table is unaffected)
+PAYLOAD_TEMPLATE = {
+    'ZSym': '',
+    'OutOpt': 'PIC',
+    'Output': 'on',
+    'NumAdd': '1',
+    'Energies': '',
+    'WindowXmin': '0.001',
+    'WindowXmax': '100000',
+    'ResizeFlag': 'on',
+    'Graph0': 'on',
+}
+
+# ==============================================================================
+# QUERY XCOM AND GENERATE MASS ATTENUATION HDF5 FILE
+
+print('Generating mass_attenuation.h5...')
+
+with h5py.File('mass_attenuation.h5', 'w') as f:
+
+    for Z in range(1, 101):
+        print(f'  Processing {ATOMIC_SYMBOL[Z]} (Z={Z})...')
+
+        # Build and submit POST request for this element
+        payload = urlencode(dict(PAYLOAD_TEMPLATE, ZNum=str(Z))).encode('utf-8')
+        req = Request(XCOM_URL, data=payload, headers={'User-Agent': 'openmc-data/1.0'})
+        with urlopen(req) as response:
+            page = response.read()
+
+        # Parse the HTML results table.
+        # Row structure: rows 0-2 are headers; data rows follow.
+        # Columns: 0=edge label, 1=energy(MeV), 2=coherent, 3=incoherent,
+        #          4=photoelectric, 5=pair(nuclear), 6=pair(electron),
+        #          7=total w/ coherent, 8=total w/o coherent
+        tree = html.fromstring(page)
+        trs = tree.xpath('//table//tr')
+
+        energies = []
+        mu_rho = []
+        for tr in trs[3:]:  # skip 3 header rows
+            tds = tr.xpath('.//td')
+            cells = [td.text_content().strip() for td in tds]
+            if len(cells) < 8:
+                continue
+            try:
+                energies.append(1e6*float(cells[1]))  # convert MeV to eV
+                mu_rho.append(float(cells[7]))  # total attenuation w/ coherent
+            except ValueError:
+                continue
+
+        if not energies:
+            raise ValueError(f'No data parsed for Z={Z}')
+
+        # Convert to numpy arrays
+        energies = np.array(energies)
+        mu_rho = np.array(mu_rho)
+
+        # Only include values up to 20 MeV
+        mask = energies <= 20e6
+        energies = energies[mask]
+        mu_rho = mu_rho[mask]
+
+        # Create dataset as 2D array: row 0 = energy (eV), row 1 = mu/rho (cm2/g)
+        data = np.array([energies, mu_rho])
+        f.create_dataset(f'{Z:03}', data=data)
+
+        # Be respectful to the NIST server
+        time.sleep(0.5)
+
+print('Done! Wrote mass_attenuation.h5')