This project used the Monte Carlo N-Particle (MCNP) code, which was developed by Los Alamos National Laboratory (LANL). MCNP code uses the Monte Carlo method to simulate the propagation of particles, including neutrons and photons (http://mcnp.lanl.gov/). To the benefit of our project, the new version of the MCNP code contains the latest cross-sectional data and is able to tally the neutron flux, activation, and radiation dose based on user-defined source/moderator/reflector/shielding geometry and composition. At thermal neutron energies, the binding of the scattering nucleus in a solid, liquid, or gas affects the cross-section and the angular and energy distributions of the scattered neutrons. When available, the S(α, β) data were included to better simulate thermal neutron interaction. This project used MCNP5 version 1.6 of the code, and all input files were checked with the VISED X_24E visual editor for geometric consistency. All the results have uncertainties of less than 5%.

The neutron generator used in this project was a customized DD-109 manufactured by Adelphi Technology Inc. (Redwood, CA). The main components of a DD neutron generator are the ion source, ion extractor, beam target, power supply/electronics rack, and heat exchanger. Figure 1 shows the DD neutron generator installed in our lab, while Figure 2 is a schematic plot of the generator head with dimensions provided by Adelphi. A deuterium (D₂) gas bottle was mounted beside the table to provide a continuous supply of deuterium gas when the system was operating. The gas line was highly vacuumed by a roughing pump and a more rigorous turbo pump. Like all DD neutron generators, the DD-109 employed in this study used the DD fusion reaction (²D + ²D → ⁴He → ³He + n) and was driven by an ion beam supplied by a radio frequency–driven ion source.

The target for this project was made from titanium-coat ed copper. To maximize the neutron production and lifetime of the target, the temperature of the titanium surface was maintained by active cooling. The V-shaped target was also designed for efficient cooling, as shown in Figure 2. To shield the bremsstrahlung x-rays generated by the electrons emitted back to the plasma source from the primary ion interaction at the titanium target, 3 mm thick lead was placed around the generator head.

Neutron flux of up to 3 × 10⁹ neutrons/second can be produced with this generator, depending on the acceleration voltage and the ion current. The voltage varies from 80 kV to 125 kV, while the current varies from 10 mA to 13 mA. The true neutron flux can be determined by a lightweight NSN3 neutron survey meter (Fuji Electric Corp.) coupled with MC simulation results. This neutron survey meter uses mixed methane and nitrogen gas to measure fast neutrons based on elastic scattering reactions and thermal neutrons based on ¹⁴N(n,p) reactions. The neutron ambient dose equivalent can then be obtained, taking into account the ICRP 74 (ICRP, 1997) neutron flux-to-dose equivalent rate conversion factors (Nunomiya et al., 2011). The NSN3 dosimeter’s mono-energetic and continuous energy response is within 50% difference from thermal to 15 MeV neutrons.

Based on the MC simulation results presented for our past work (Liu et al., 2013), an optimized moderator/reflector/shielding system was built to create a cavity for the irradiation of human hands. Our previous paper showed that an optimized configuration consists of 5 cm of paraffin as the moderator and >10 cm of paraffin as the reflector. In the system used in the current study, polyethylene was used instead of paraffin because paraffin is flammable. This will not significantly impact the results, however, since the composition of paraffin and polyethylene are similar, as shown in our paper cited above. Specifically, 5 cm of polyethylene was used as the moderator; 10 cm of polyethylene was used as the reflector; and the shielding structure was made of >30 cm of polyethylene. The neutron dose outside of the shielding was measured to be 2–5 mR/hour based on this configuration. Figure 3 shows the experimental setup of the DD neutron generator with the polyethylene moderator/reflector/shielding system, with part of the shielding removed to present a better view of the human hand cavity.

Work is in progress to build a more compact shielding structure with a tighter fit around the generator head except on the side where the hand will be irradiated. Our most recent experimental and simulation results (not presented in this paper) also show that the neutron spectra are significantly altered as the size of the gaps between the moderator blocks and reflector blocks changes. In addition, graphite has been demonstrated to be a more efficient material for a reflector. More work continues to be conducted to further optimize this moderator and reflector configuration.

Five Mn-doped hand phantoms were manufactured and used in this study. The Mn concentrations in the phantoms were 0, 5, 10, 15, and 20 μg Mn/g bone (which corresponds to 0, 22, 44, 66, and 88μg Mn/g Ca). Other elements in bone that might interfere with the spectrum through neutron activation were also added to the phantoms to better simulate real human hands. The concentration of each element in the bones of the hand was calculated based on ICRP publication 23’s gross and element content of the cortical bone of a reference human male (ICRP, 1975). Table 1 lists all the elements included and the weight of their chemical compounds.

All the chemical compounds were first diluted in distilled water before being added to the matrix to ensure a better homogeneity of the elements in the phantoms. The phantoms were then dried in the hood for one day. These phantoms were bone-equivalent phantoms, but they did not have the shape of a human hand. More work is in progress to manufacture hand-shaped phantoms encased in soft tissue.

With the system shown in Figure 3, Mn concentrations present in the hand bone of a human subject can be noninvasively determined using in vivo neutron activation analysis (IVNAA). As described in our previous paper (Liu et al., 2013), during neutron activation, characteristic γ-rays are produced following the radioactive decay of the product from an ^AX(n,γ)^(A+1)X nuclear reaction. By collecting the characteristic γ-rays and calculating their total counts, the concentration of the element of interest can be determined. For Mn quantification, neutrons interact with ⁵⁵Mn and produce ⁵⁶Mn with a thermal neutron capture cross section of 13.3 barns. Unstable ⁵⁶Mn atoms decay to ⁵⁶Fe, which emits 847 keV characteristic γ-rays. These γ-rays can then be collected by a γ-ray detection system. ⁵⁶Mn’s relatively long half-life of 2.58 hours allows for delayed γ counting. The calculation for the intensity of the characteristic γ-rays can be found in our previous paper.

Using the fm4 card in MCNP5, the probability of the activated nucleus can be obtained. Together with the activation equation, the simulated total γ-ray counts can be expressed as: (1)CTotal=N×γ×ε×S×D×C where C_Total is the γ-ray counts that will be measured by the γ-ray detector; N is the total activated ⁵⁶Mn number from the simulation result; γ is the branch ratio of the γ-rays; ε is the absolute detection efficiency; S(= 1 − e^−λt_i) is the saturation factor with t_i representing irradiation time; D(= e^−λt_d) is the decay factor with t_d representing decay time; and C(=1 − e^−λt_c)/λ is the counting factor with t_c representing counting time. C_Total can be compared to experiment results to test the consistency of the simulation and experimental results.

The irradiation, decay, and counting time can also be optimized to determine the best time sequence. We selected 10 minutes of irradiation time to allow for an acceptable dose to the hand; 5 minutes of decay time to collect a spectrum for calcium (Ca); and 30 minutes of measurement time in consideration of the time that a human subject could be expected to sit relatively still to take the measurement. This time sequence can be further improved in future work.

A sample of pure gold (Au) foil and an Mn-doped hand phantom were irradiated by the DD–based neutron generator system and then measured by an HPGe γ-ray detection system. The same scenarios were also simulated using the MC simulation model. The results from the MC simulations and the experiments were then compared. This is to validate the results from MC simulations.

A high-efficiency HPGe detector was used in this study for γ-ray detection. It is a model GMX90P4-ST HPGe detector with a relative efficiency of 100%. The detector is cooled by an electromechanical cooler (Ortec, Oak Ridge, TN). Lead bricks were mounted around the detector to reduce the background signal. The DSPEC Plus digital box was used for signal processing, and Maestro γ-ray spectroscopy software was used for signal collection. The efficiency of the system was calibrated using a multi-radionuclide calibration source with known activities. The efficiency equation was obtained as: efficiency =1.0548 × energy(keV)^−0.688 at 5 cm away from the detector’s window. Figure 4 shows two high-efficiency detectors with a possible configuration of how a hand could be measured. Notably, only one of these detectors was used in this project because the other one was experiencing some technical problems.

Gamma-ray spectrum analysis was performed using an in-house fitting procedure programmed in the commercial software package IGOR Pro 6 (Wave Metrics, Inc., Lake Oswego, OR). The γ-ray peaks were fitted using the Gaussian function for net counts and an exponential function to account for background.

The most straightforward way to calibrate the system for Mn quantification is to build a calibration line of ⁵⁶Mn γ-ray counts versus Mn concentration. However, this count would be affected by the thermalization of the neutrons within the samples, the thickness of the soft tissue in the hand, the weight of the hand, and the slightly different irradiation geometries. To account for these differences, Mn γ-ray counts can be normalized to Ca γ-ray counts, since the concentration of Ca is relatively constant in bone. Thus, a calibration line representing the Mn/Ca ratio versus Mn concentration was established.

As described above, neutron flux can be determined using a neutron generator coupled with MC simulation results. Figure 5 shows a cross-sectional plot of the neutron activation system used in the MC simulation model. To produce approximately 10⁹ neutrons per second, the magnetron was set to 5 kV and 80 mA; the accelerator was set to 120 kV and 16 mA; and the D₂ gas flow was set to 1.2 standard cubic centimeters per minute (SCCM) with a pressure of ~5.0 mTorr (~4.7 mTorr in operation). The neutron dose, as measured from one side by the NSN3 dosimeter when the shielding was open, was 948.6 mRem/h at 109 cm away from the neutron source. The NSN3 neutron survey meter measures ambient neutron dose equivalent H*(10), so ICRP-74 (ICRP, 1997) neutron flux-to-dose equivalent rate conversion factors were employed in MCNP5 input files to obtain a dose measure that equaled the ambient dose equivalent. The simulated neutron dose for the neutron flux of 1 × 10⁹ neutrons/second at the 109 cm was 1374.5 mRem/h. Given the measured dose of 948.6 mRem/h at the same spot, the actual neutron flux was calculated to be about 7 × 10⁸ neutrons/second.

With the parameters and settings described above, and in order to assess how the MC simulation results compared to experimental results, a 0.121 gram sample of Au foil was irradiated in the irradiation cavity for 10 minutes, decayed for 2 hours, and measured for 1 hour with the HPGe γ-ray detection system. Additionally, a 20 ppm Mn-doped hand phantom was also irradiated in the irradiation cavity for 10 minutes, decayed for 10 minutes, and measured for 30 minutes with the HPGe detection system.

The Au γ-ray counts calculated from the experiment was 818.2±27.6, compared to the simulation result of 809.4±1.1. The Mn γ-ray counts calculated from the experiment, based on the net peak counts from the spectrum analysis was 238.8±24.8, compared to the simulation result of 141.2±0.2 taking into account the detector efficiency. Several factors might have contributed to the discrepancy for Mn measurement and simulation, with the main factor to be the difficulty to determine the detector efficiency for the measurement of the hand phantom (in contrast, it is much easier to obtain an accurate detector efficiency for the measurement of Au foil); these factors are considered in the discussion section.

Mn-doped hand phantoms with Mn concentrations of 0, 5, 10, 15, and 20 ppm were placed in the sample cavity and irradiated for 10 minutes, decayed for 5 minutes, and measured by the HPGe detector for 30 minutes. The spectrum collected from the 5 ppm phantom is illustrated in Figure 6. With an energy resolution of 2.0 keV at 1.33 MeV, the peak of Mn γ-ray at 847 keV can be clearly seen. The background in that energy range is mainly from Compton scattering and is relatively low. An enlarged Mn peak is shown at the top right of the figure. A magnesium (Mg) γ-ray peak at 844 keV is also observed from the spectrum. Although the Mg concentration in the hand phantoms is much greater than the Mn concentration, as shown in Table 1, the interference of the Mg γ-ray peak is minimal, even with a Mn concentration at as low as 5 ppm, as shown in Figure 6.

The detection limit (DL) of the system was calculated based on the measurements taken from the Mn-doped phantoms. It was calculated using the following formula: (2)DL=2×backgroundC where background is the background counts under the Mn γ-ray peak for the 0 ppm phantom, and C (counts/ppm) is the slope of the regression line of Mn counts versus Mn concentration. The energy range of the background was estimated to be 4 sigma of the Mn γ-ray Gaussian peak, which covers 96% of the peak counts. Sigma was estimated from the fitting program built into IGOR. The DL calculated from equation 2 was 1.05 ppm.

The spectra for all the Mn-doped hand phantoms (0, 5, 10, 15, and 20 ppm) were analyzed. The net peak counts for Mn and Ca were calculated using our in-house peak fitting program. The Mn/Ca ratio versus Mn concentration was then plotted, as shown in Figure 7. The R-square for the correlation was 0.99, which indicates that the phantoms are a good configuration and that the calibration procedure works.

The radiation dose to a human hand was calculated using MC simulations. After 10 minutes of irradiation at a neutron flux of 7 × 10⁸ neutrons/second, the equivalent dose to a hand was found to be 85.4 mSv. The dose outside of the shielding was calculated using MC simulations and measured using the NSN3 neutron meter. The simulated dose rate was 3.88 mRem/hour or 38.8 μSv/hour with the neutron flux of 7 × 10⁸ neutrons/second, which gives rise to a dose of about 6.5 μSv for 10 minutes of irradiation. The measured dose rate was about 3.85 mRem/hour or 38.5 μSv/hour, which is very close to the simulated value. The weight of a human hand accounted for about 1.25% of the weight of the whole body. The tissue weighting factor for the more sensitive organs in hand, skin and bone surface, is 0.01. Taking into account the tissue weighting factor, the whole body effective dose was calculated to be 85.4×10³×0.01×0.0125+6.5 = 17 μSv. For comparison, the whole body effective dose from a standard AP chest x-ray is about 100 μSv. The neutron spectrum inside the hand irradiation cavity is complex, and work is being conducted to obtain an accurate neutron dose inside the hand cavity using experiments to validate the hand dose calculated by simulation.