Nucl. Tracks, Vol. 10. Nos 4-6, pp. 737-742, 1985 0191-278X/85 $3.00+0.00 Printed in Great Britain Pergamon Press Ltd.

TL DATING OF LOESS AND FOSSIL SOILS FROM THE LAST INTERGLACIAL-GLACIAL CYCLE

HANNA PR6SZY~ISKA-BORDAS

Department of Geography and Regional Sciences, University of Warsaw, Krakowskie Przedm. 30, 00-950 Warsaw. Poland

(Received 30 November 1984; in revised fornr 23 January 1985)

Abstraet--TL dates on fine grains from loess and fossil soils have been obtained using the near ultraviolet emission of feldspars. The results are compared with data obtained previously with a blue filter, which enhances the emission of quartz. The values of equivalent doses calculated around 300°C using the "regen" method are in good agreement for both filters. For samples less than 50 ka the plateau test is particularly 'well fulfilled when using the near ultraviolet emission. Further down the section good plateaux were not obtained and the absolute meaning of older TL ages is still under question. In spite of this, the apparent TL ages provide enough evidence to solve particular chronostratigraphic problems, for example, enabling us to distinguish two periods of interglacial soil formation, which have been miscorrelated up to now.

1. INTRODUCTION

AT THE last TL dating seminar in Denmark I presented some TL ages of loess (Proszyfiska, 1983). In that study, glow curves were taken using the Corning 5-58 blue filter and the "regen" method (regeneration of the TL after long bleaching) was applied. Compared with the "addit ive" methods, where the laboratory TL is superimposed on top of the natural, the "regen" method yielded much better plateaux (Wintle and Prrszyfiska, 1983). However, the large residual signal in young sediments and the lack of a long plateau in older ones caused problems with the age determination.

Instead of the traditionally preferred blue emission Debenham and Walton (1983) proposed the use of the u.v. signal of feldspars, which is more readily bleached and gives linear growth curves to higher doses than polymineral blue emission which includes the quartz signal. This method seemed promising, but the latest reports suggest that when applied to loess its time range will not go much beyond 100ka (Debenham, 1985; Wintle, 1985). It would be inter- esting to know if the same happens in the blue region, so the main aim of this paper is to compare the results obtained for each wavelength region.

2. M E T H O D

Most of this work was carried out in the TL laboratory of Dr Ann Wintle in Cambridge. The

method of the equivalent dose (ED) evaluation was the same as described by Wintle et al. (1984). The' beta-irradiations were performed on discs previously bleached during 500 min under the mercury lamp. The a-value was assessed by applying an alpha dose to bleached samples. To get rid of the low tempera- ture TL each disc was preheated at 230°C for 1 min before the main run. The near u.v. signal was detected with a Schott UG-11 filter used together with a Chance-Pilkington HA-3 heat absorbing filter, the resultant spectral region being 300°380 nm. The EDs were usually calculated in the temperature region 250-350°C.

The dose-rates were evaluated by s-counting and/or by 3'-spectrometry (Prrszyfiska-Bordas, 1983). The cosmic dose contribution was assessed according to the geological situation and the mean water content A of 20 + 5% was assumed for the upper, actually dried up, part of the exposure.

3. SAMPLES AND RESULTS

A large number of samples from different localities has been examined layer by layer. As an example TL results obtained for a typical loess profile in Lopatki (51: 19' 54" N, 22: 09' 00" E, Lublin Upland, Poland) are presented in Fig. l(a) and Table 1. In Fig. l(b) EDs are plotted for the samples analysed with each filter. The results obtained with the 5-58 filter are for loess sites at Krakrw Zwierzyniec and Kazimierza

737

738 H. PROSZYlqSKA-BORDAS

Wielka-Odon6w (see Table 1 in Pr6szyfiska, 1983). The near u.v. results are mainly for Lopatki, the two oldest samples come from longer profile at Nieledew. Strikingly similar values of EDs have been obtained for at least five other profiles, enabling to distinguish several age groups, clearly recognizable in Fig. l(b).

The ratio of the TL left after bleaching to the total natural TL is almost a factor of two smaller for the UG-11 filter (Table 2).

As a result of pre-heating at 230°C for 1 min, the glow curves e x h i b i t a maximum between 295 and 305°C (heating rate 5°C s-~). This maximum falls at the same temperature for both natural and regen- erated discs only in groups I - I I I . For older samples, the pre-heating does not lead to the exact fitting of natural and regenerated glow curves, the regenerated ones having maxima 2-3°C lower, even 5°C if very large doses were applied.

Loess samples from groups I and II have EDs below 100 Gy and do not appear to have problems. Use of either filter leads to good plateaux. If we perform the measurements on differently bleached subsamples we can choose the residual level charac- terised by the best plateau (Pr6szyfiska-Bordas, 1983; Mejdahl, 1985). For loess the best plateaux are usually obtained when the laboratory exposure to light is long enough to remove almost all bleachable TL. The youngest group (I), typically having EDs of 50-65 Gy, represents the final part of Last Glacial with the TL ages around 15 ka before present. The next group (II) includes loess deposited on top of an interstadial arctic soil, with EDs of 80-90 Gy and TL ages of some 21-23 ka.

Group III contains samples taken between the above-mentioned interstadial soil and a well devel- oped complex soil of interglacial range. Here we get good plateaux using the UG-I 1 filter and generally poorer ones using the 5-58. The EDs obtained around 300°C are almost identical in all investigated sections--in Eopatki 130 Gy at the top of the stratum and 160 Gy at the bottom. This yields ages between 50 and 30 ka.

The samples from the so-called Last Interglacial pedocomplex were analyzed only with the U G - I I filter, good plateaux being found for upper humus accumulation horizons (At) described as a steppe or meadow chernozem soil, which was formed on top of a forest soil by the end of the interglacial. The ED of 160 Gy gives a TL age of 50 ka, which falls much later than the end of the Last Interglacial. However, the

dated event was possibly the burial of the soil: bioturbation of this organic-rich horizon could have allowed it to have been bleached until it was covered by new sediment. In no case is its age older than 70-60 ka, which is the same as for the next lower horizon A 3, eluvium, which genetically belongs to the forest soil, presumably decapitated by erosion. In my

• opinion the event dated in the case of A 3 is the end of erosion, when the eluvial horizon was at the surface and subject to turbation. Both A~ and A3 horizons were undoubtedly enriched in wind-blown silt and deluvial material, which gets bleached during its migration down the slope. Although we still know very little about the bleaching of TL in soil, good plateaux and consistency of TL results between several sites enables us to conclude that bleaching in A~ and A 3 horizons of this pedocomplex was sufficient and occurred in distant localities at the same time in the past.

Analysing the still lower Bt soil horizon, illuvium, we face the question "what was the event registered by TL in this case, deposition of parent loess or later soil formation?". Although the B t horizon is enriched in colloidal substances leached in from upper soil horizonL the TL of silt-sized grains should remain unaffected. The B t horizon of modern soils has a TL level corresponding to the geological age of the parent sediment. There is also a problem of possible loss of TL due to weathering of minerals. Post- depositional weather would have affected the feld- spars to a greater extent than the much more resistant quartz. The similarity of the EDs obtained with the 5-58 and UG-11 filters indicates that either weather- ing is unimportant or it affects both emissions to the same extent. In decalcified loess, where the weather- ing indexes are high, the problem may be more serious.

Samples from the lowest part of the illuvium and from parent loess underlying the soil form group IV. Here the EDs calculated in the same temperature region as for younger groups are around 300Gy yielding a TL age of 80 ka. However, in almost all samples, even when the UG-I 1 filter was used, the EDs rise slightly with temperature: the ratio of the ED at 400°C to that at 300°C is 1.19 + 0.06 (data for five sites). However, for even older sediments the ratio is usually not larger than this, indicating that thermal decay of the 300~C TL is not the sole reason for the poor plateaux. Using the UG-I I filter in sediments further down the section we still do not

TL D A T I N G O F LOESS A N D FOSSIL SOILS

A B age t • m p e r a t u r e ( ° C )

I k . 1 % 2~o 21o 3oo ~;o ~ o

.so , , , . . . . , , , , , ~ ~ t ] d loess I ~ t t 0 0 D 0 0 0 0 D 0 O 0

3 6 t 4 .I00

i n t e r s t a d i a l a r c t i c s o i l 38*--4

D A

v l o e s s ! I I

" ~ T !

,,,,,t

1 o e s s"~l V t I

80±9

:;::: t ttttttt t t h e o l d e r

, ~ - N 3 p e d o c o m p l e x

FIG. 1.(A) Simplified scheme of lit"hological sequence in Eopatki with sample positions and TL ages. (I) Interglacial soil with genetic horizons: (A0 accumulation, (A3) eluvium, (B,) illuvium; (2) interstadial soil with fossil ice-wedges forms; (3) loess; (4) gleyed loess; (5) sand; (6) glacial till. (B) Equivalent dose plots (caution: ED grows downward, plateaux reversed!) for typical samples from each age group distinguished so far in loess sections in Poland: V, results for near u.v. emission with + lu error bars; I"l, results for blue emission after Proszyfiska (1983) for stratigraphically analogous sam~es from Krak6w-Zwierzyniec

(Ist. 2nd and 4th plot from the top) and Kazimierza Wielka-Odon6w (3rd and 5th plot).

739

740 H. PROSZYlqSKA-BORDAS

6 0 k

o o

\ ,e

0

4 0 k 0

. d

2 0 k

N a t u r a I

/

/

/

/

/

/

/

T L • /

/ •

• /

/

/

/

/

/

• / • / o o /

/

/

/e /

/

i I i J i i i ,

200 400 600 800

B • t a d o s e ( G y )

FIG. 2. Growth curve (temperature 290-299°C) for the loess horizon Niel n from Nieledew, where a palaeomagnetic reversal was found; dashed line is a linear least squares fit using all points.

observe saturation; whereas the regenerated growth curve is yet rising, the natural TL does not exceed a certain level, never above 1000 Gy (Fig. 2). Such an effect was also observed with both filters in very old samples, believed to date back to the Brunhes/Matuyama and Matuyama/Gauss palaeo- magnetic boundaries (Table 3). Here the growth curves are considerably non-linear. The TL levels after standard lamp bleaching were still very high, so the original residual level as for loess group IV was assessed and EDs calculated according to this. The comparison of TL sensitivity before and after bleach- ing, needed to verify if the underestimation of TL results is caused by the increase of sensitivity, was not yet done for these particular samples. However, the other TL studies on loess indicate that there is no general tendency of significant sensitivity increase due to the bleaching in both blue and u.v. regions (Wintle and Pr6szyfiska, 1983; Debenham, 1985).

4. CONCLUSIONS

The data obtained with the Schott UG-I 1 and the Corning 5-58 filters are in good agreement and enable direct correlation between both sets of TL ages. The UG- 11 filter is particularly useful for dating loess less than 50 ka old, for which very good plateaux have been obtained (starting as early as 200°C due to the pre-heating). In older samples a rise in ED with temperature is common, but still the apparent TL age can be valuable for chronostratigraphic correlation within the Last Interglacial-Last Glacial climatic cycle (125-12 ka).

As in loess dating projects in other parts of Europe it can be seen that the EDs and resulting TL ages are concentrated into a few groups. Not only does this change the traditional views that a particular loess section represents much of the Pleistocene history, but it also has some implications in the field of

TL D A T I N G O F LOESS A N D FOSSIL SOILS

Table I. TL results and radioactivity data

741

Depth ED ~, Count-rate U (~) Th (~) Sample (m) (Gy) a-Value (ks- l cm- 2) (p.p.m.) (p.p.m,)

Lop-d loess I 2.10 50 _+ 2 0.110 Lop-e~ loess II 2.25 84 4- 5 0.103 Lop-e: arctic soil on loess III 2.50 133 4- 3 0.108 Lop-f~ loess III top part 2.90 133 _+ 3 0.107 Lop-i 2 loess III bottom part 7.05 161 4- 3 n.m. 0.630 2.83 8.3 Lop-j~ chernozem top part 7.35 160 4- 5 0.099 0.612 3.29 6.2 Lop-j2 chernozem bottom part 7.55 174 4- 6 n.m. 0.645 3.61 6.1 Lop-k eluvium 7.65 240 4- 7 0.115 Lop-14 illuvium on loess IV 8.70 264 4- 7 0.113 0.595 3.07 6.5

Notes: (:t) Thick sample ~t-counting results. (~) ?-spectrometry results. Dose-rates calculated for averaged U and Th.

Table 1. (cont)

U (r) Th (~) K20 (~) Dose-rate TL age (p.p.m.) (p.p.m.) (%) A (mGy a - l) (ka)

2.68 7.35 1.71 0.20 3.28 15 + 2 n.m. n.m. n.m. 0.20 ~ 23 2.47 9.51 1.87 0.20 3.65 36 4- 4 3.05 7.08 1.94 0.20 3.52 38 4- 4 2.63 8.68 2.09 0.21 3.62 44 4- 5 3.17 7.59 1.80 0.24 3.24 494-5 3.02 7.07 1.87 0.26 3.25 53 + 5 3.45 7.86 1.92 0.17 3.89 62 4- 7 2.33 6.27 1.78 0.17 3.30 80 4- 9

dosimetry and TL dating theory. If for .dis tant sites we get the same sequence of EDs and ages, it will mean that the natural system of radioactive emitters and fine silt dosimeters in loess do not much depend on different local condit ions (water content, mi- gration of radionuclides, mean tempertures, etc.). Also, the wind-blown silt grains must have been well bleached before burial.

We cannot consider the geologists ' expected ages to provide a check of stability o f TL, because the geological time-scales are sometimes based on very loose correlations. On the other hand, we should look for reasons for the age underest imation encountered in older sediments, checking both the possibility of loss of TL in situ and the adequacy of laboratory procedures. Tests of the mean life of a particular TL signal are necessary. There is still a long way to go before we can define the limits o f TL dating of loess and establish their causes.

5. A P P L I C A T I O N S

Geochronological implications are sometimes far from tradit ional views. One is especially striking: on the basis o f the TL results we are able to distinguish at least two "Las t Interglacial" soils, up to now thought to be of the same age. The older one is sandwiched by layers having EDs almost twice as

Table 2. Typical ratios of TL left after bleaching to the total natural TL at 290°C

Near u.v. filter Blue filter Loess group Schott UG-11 Coming 5-58

I 0.15 0.25 II 0.08 0.15 III 0.06 0.13 IV 0.04 0.09

742 H. P R O S Z Y l q S K A - B O R D A S

Table 3. Equivalent doses calculated for the TL peak of 300~C after pre-heating at 230:C for 1 min

ED (Gy) ED (Gy) Near u.v. filter Blue filter

Sample Schott UG-II Corning 5-58

Paks (Hungary) 990 ___ 80 soil from the boundary Brunhes/Matuyama 0.7 Ma

Stranzendorf (Austria) 760 + 50 loess from the boundary Matuyama/Gauss 2.5 Ma

860 + 40

750 __+ 30

large as the loess under- and overlying the younger soil. The older soil hase been detected in Saint Romain (France) by Wintle et al. (1984), who cor- relate it with the climatic opt imum of the Last Interglacial (oxygen substage 5e, about 120 ka). If it is true, the formation of the younger soil could possibly have started in one of subsequent warm substages 5c or 5a, dated on Barbados by the 23°Th/234U method as 105 and 82 ka (Mesolella et al.,

1969; Shackleton and Matthews, 1977). The palaeo- magnetic evidence does not contradict such a par- tition. Above the older soil a palaeomagnetic reversal correlated with the Blake event has been found in Poland (profiles Komar6w G6rny and Orzechowce) by Tuchotka (1977). No anomaly could be traced either inside or above the younger soil complex in Kazimierza Wielka-Odon6w and Nieledew, however, in Nieledew it was found above the second from the top pedocomplex (Tuchoika, 1977). No reversal was detected in Kopatki and judging from the registered changes of magnetic field the active deposit ion of silt lasted only a few thousand years (M. Tkacz, personal communicat ion) . This agrees with results presented

here.

Acknowledgements--I am indebted to Dr Ann Wintle for including me in the TL grant from NERC (GR3/3174) and for her help during my stay in the Godwin Laboratory. 1 thank Professor H. Maruszczak who showed me the loess

outcrops. The volunteer help of Mrs W. Stafiska- Pr6szyfiska, who operated the y-spectrometer, is particu- larly acknowledged.

REFERENCES

Debenham N. N. (1985) Use of u.v. emissions in TL dating of sediments. Nucl. Tracks 10, 717-724.

Debenham N. C. and Walton A. J. (1983) TL properties of some wind-blown sediments. PACT 9, 531-538.

Mejdahl V. (1985) Thermoluminescence dating of partially bleached sediments. NucL Tracks 10, 711-715.

Mesolella K. J., Matthews R. K., Broecker W. S. and Thurber D. L. (1969) The astronomical theory of climatic change, Barbados data. J. Geol. 77, 250-274.

Pr6szyfiska H. (1983) TL dating of some subaerial sedi- ments from Poland. PACT 9, 539-546.

Pr6szyfiska-Bordas H. (1983) Termoluminescencyjne ws- ka~,niki wieku osad6w jako narz~dzie rekonstrukcji paleogeograficznej. Unpublished D Phil. Thesis, Uni- versity of Warsaw.

Shackleton N. J. and Matthews R. K. (1977) Oxygen isotope stratigraphy of Late Pleistocene coral terraces in Barbados. Nature, Lond. 268, 618-620.

Tuchotka P. (1977) Magnetic polarity events in Polish loess profiles. BiuL Inst. Geol. 305, 117-123.

Wintle A. G. (1985) Stability of TL signal in fine grains from loess. Nucl. Tracks 10, 725-730.

Wintle A. G., Shackleton N. J. and Lautridou J. P. (1984) Thermoluminescence dating of periods of loess deposi- tion and soil formation in Normandy. Nature, Lond. 310, 491-493.

Wintle A. G. and Pr6szyfiska H. (1983) TL dating of loess in Germany and Poland. PACT 9, 547-554.