TGA
Mass loss occurs over three distinct temperature ranges (Fig. 2).
First, when heated from 25 to 180⁰C, the sample loses about 8 wt%, due to loss of H2O from various sites including collagen water of hydration, and water trapped in voids in mineral. From about 300 to 550⁰C, the sample loses about 27 wt% by the burning of collagen and other organic components. The minimum in the derivative curve at ~560⁰C suggests that burning of organic matters has ended. Between 550 and 900⁰C there is an additional loss of about 4.1 wt% presumably due to decomposition of CO32- in apatite, liberating CO2. The pattern shown here is very similar to that shown by Greiner et al. [13]. We can generally recognize three distinct stages in weight loss, separated by points at which the rate of mass loss goes through minima (red vertical lines on Fig. 2).
Constancy of furnace temperature
When a sample is inserted into the furnace, the temperature immediately falls below the nominal temperature to which the furnace was set, which was displayed on the thermometer before insertion. After closing the furnace door, T slowly returns to the nominal temperature (Fig. 3).
Preheating
Preheating was carried out at 60⁰C for 30 min or 24 h. The average weight loss of liquid and adsorbed water for samples ranging in weight from 10 to 100 mg was 7.8± 0.8 wt% for both preheating times.
Testing effects of variation in conditions of ashing
Samples were ashed at 500, 600 and 700°C. Samples of both larger weights, between 50 and 100 mg, and smaller weights between 5 and 20 mg were tested. Ashing times of one hour and 24 hours were used. Preheating at 60°C was also tested for 30 minutes and 24 hours.
The data obtained using these various methods are given in Supplement 2, and are summarized in Figs. 4-7.
From Fig. 4, we immediately see the nature of the problem of obtaining a single ash weight from a bone sample. As the temperature of ashing increases, the weight of the residue gradually decreases, lowering from 69.4 ±1.1 to 64.3 ±1.1 wt%. This is a 7.3% increase in weight loss. Which of these weights is the “correct” ash weight?
The heating experiment was repeated using sample weights of 5 to 20 mg to test the reproducibility of the method using smaller samples. The results are shown in Fig. 5.
We again see a drop in ash weight of 7.1% as the ashing temperature rises from 500 to 700⁰C. Some of this loss could be due to continued burning of organic material which appears to end at ~560⁰C.
The time of heating the sample would presumably also have some effect on ash weight; to test this we increased both the preheating and ashing periods to 24h (Table S2.3). The longer preheating and ashing times resulted in an average wt% ash of 66.2±0.55 while preheating for 30 min and ashing for 1 hour had given a significantly larger ash wt% of 68.0 ± 0.9 (Table S2.1).
Standard method of ashing
Based on the prior results, it was apparent that the ash weight result would depend critically on the time and temperature of heating. We attempted to develop a standard method for determining the ash weight of bone which would achieve the following: a) minimize dispersion of the data (reproducibility); b) minimize the amount of sample needed for analysis; and c) minimize time needed for the analyses. Based on our experimental runs, we selected four possible sets of conditions (Tables S2.4 and S2.5). Fig. 6 displays the results of the experiments with larger weights of bone, while Fig. 7 displays the results with smaller weights of bone.
In principle, either the 600⁰C or 700⁰C ash weights could be used as the “reference method” for determination of ash weight. For reasons presented in the “Discussion,” we have selected the smaller weights of samples and heating at 600⁰C as the standard. The complete method would be as follows:
Sample weight: 5 - 20 mg
Preheat treatment: 60⁰C for 24h, followed immediately by
Ashing: at 600⁰C for 1 h
The average obtained for our standard bovine sample using these conditions was 70.5 ± 1.1 wt%; individual percent ash weights ranging from 68-73% (Table S2.5, Fig. 7).
Human samples
In order to test the efficacy of this method, it was applied to a series of distinct human bone samples from older women and one young man. The data are summarized in Table S2.6 and Fig. 8.
Table 2: p-values for statistically significant samples shown in Fig. 8
Samples Compared
|
P- value
|
161569 vs. all other samples
|
<0.0001
|
172202 vs. Bovine bone
|
0.0003
|
171519 vs. 1553
|
0.0261
|
1553 vs. 1493
|
0.0051
|
1553 vs. 1640
|
0.0065
|
1553 vs. Bovine bone
|
<0.0001
|
19 yr vs. Bovine bone
|
0.0242
|