Document seems to be null but I can get the first element

I build a DOM document with this Code:

private org.w3c.dom.Document createXml(WorkObject workObject, String hierarchy, String dateFormat) {
    XMLBuilder2 builder = XMLBuilder2.create(hierarchy);
    Map<String, Object> localVariables = workObject.getLocalVariablesMap();
    builder = createXmlFromMap(localVariables, builder, dateFormat);
    Properties outputProperties = new Properties();
    outputProperties.put(javax.xml.transform.OutputKeys.OMIT_XML_DECLARATION, "no");
    String xmlString = builder.asString(outputProperties);
    LOG.debug("XML string: " + xmlString);
    org.w3c.dom.Document xmlDoc = builder.getDocument();
    LOG.debug("DOM root element: " + xmlDoc.getDocumentElement().getTagName());
    return xmlDoc;
}

the builder.asString()call returns the expected XML:

<?xml version="1.0" encoding="UTF-8"?>
<root>
    <myVar>a value</myVar>
    <docId>GEAR-b3384419-36f8-48fc-b194-c937e7afc0dd</docId>
</root>

The output of the last LOG statement is root as expected.

But when I use this code, the xmlDoc seems to be something weird...

when passing the xmlDoc to this constructor:

private org.w3c.dom.Document mXmlDoc;

public XmlMailMergeDataSet(org.w3c.dom.Document xmlDoc) {
    mXmlDoc = xmlDoc;
}

When debugging in IDEA, the value of xmlDoc and mXmlDoc then is [#document: null]

Unfortunaly, I can not debug further to see what happens then since the code where the XmlMailMergeDataSet then is passed to is not open source.

Am I missing something obvious here?