This depends on your samples. For some quartz grains, these doses are OK. Signals from other samples may be close to saturation. In this case, you have to analyze the shape of the growth curves as mentioned above, and try to find indepentent age controls to determine the reliablility of the ages obtained.

Let me first ask one question, what mineral you have used? Quartz, feldspar or something else.

first of all thank you for your response. we use quartz for dating. the size of grains is from 100-250 micrometers and we follow the usual chemical treatment, HCl, H2O2, HF and water washings.

That depends on the shape of your dose response curve. Usually the lower saturation limit for quartz OSL is 150 to 200 Gy. A good read is the OSL review by Wintle&Murray 2006.

Tony is correct about the behavior of quartz, it has two plateau in dose response curve. one can have a good estimate of ages for a quartz of dose upto 100 Gy after which it shows saturation. However people have reported no saturation and a dose recovery test within 98% accuracy upto 3000 Gy. This has been done by LMOSL and slow component has been used, which has been reported saturating at higher doses as compared to fast components.e.g. see Timar-Gabor et al. 2010 in QI page 62-70 and Timar-Gabor and Wintle 2013 QG page 34-40.

Usually if you're using exponential plus linear fits, it is usually best to not use data on the 'linear part' of the fit. If you are using the linear part, your data may be less reliable.

If you're just using the (rare in most samples) grains that do not saturate or go 'linear' at your 400-500 Gy dose, there's no theoretical reason why your result should be unreliable, apart from the fact that your sample size will be small.

If you are measuring a signal that does not come from the fast component OSL trap (eg, slow components, TT-OSL), make sure you are aware of the lifetime of the signal you are measuring. Some have lower lifetimes than the normally-measured OSL signal, and can give unreliable results for older samples.

We do use the results on the "linear" part of the exp+lin fit of the growth curve. So as you all say that it is not reliable to use such results, what can we do to improve to reliability of the data? We also perfom at the same disks (used to produce the growth curve plus the natural signal) dose recovery test at the same order of given dose (200 Gy) and we have very good results (100-105%). The preheat is 240C and we also perform IR test where the signal is quite low if it is compared with the OSL signal (

This is true even one can use the linear part in exponential fitting or lin+exp fitting. The models are such that when you are in linear region the parameters for the estimation of dose used are mainly from the linear side of the model. besides you can test (Moffatt) that if the data is linear and you try to fit saturating exponential the estimated parameters will try to give you the linearity in the fit. Therefore to best of my understanding and practice I do not see a major problem. The only problem in saturating exponential fits is that the statistical error and interpolation errors are increased. Which partially supports the statement given by Moffatt.

This depends on your samples. For some quartz grains, these doses are OK. Signals from other samples may be close to saturation. In this case, you have to analyze the shape of the growth curves as mentioned above, and try to find indepentent age controls to determine the reliablility of the ages obtained.

Reading wintle and murray (2006) I recognised that the analysis of growth curves covers the case where one or two exponential decay fits are used and there the Do quantity is established as the dose that produces the (1-1/e) fraction of the Lx/Tx at saturation. And after that they conclude that De must be

Try using the TT-OSL signal. There are several measurement protocols, all are ok. TT-OSL is not as thermally stable as the OSL signal, but if you get the same De as for OSL then you can assume that your results are reliable. I can send you some relevant references..

Dear Naomi!

I would appriciate if you could send me these references for TT-OSL.

Thak you all for your answers.